From 794fed59b720a4d326234cb1a1e1922e23fae7e2 Mon Sep 17 00:00:00 2001 From: Krystian Igras Date: Thu, 5 Sep 2019 01:16:33 +0200 Subject: [PATCH] Cran release fixes. --- DESCRIPTION | 4 +- NEWS.md | 8 + R/methods-xspliner.R | 54 +-- R/utils-formula-build.R | 2 +- cran-comments.md | 51 +-- docs/articles/automation.html | 358 +++++++++--------- docs/articles/cases.html | 58 +-- docs/articles/discrete.html | 56 +-- .../figure-html/unnamed-chunk-5-1.png | Bin 89413 -> 93825 bytes .../figure-html/unnamed-chunk-9-1.png | Bin 44062 -> 43505 bytes docs/articles/extras.html | 156 ++++---- docs/articles/graphics.html | 134 +++---- docs/articles/methods.html | 186 +++++---- .../figure-html/unnamed-chunk-7-1.png | Bin 21587 -> 21606 bytes docs/articles/xspliner.html | 132 +++---- docs/index.html | 4 +- docs/news/index.html | 20 +- docs/pkgdown.yml | 2 +- docs/reference/build_xspliner.html | 6 +- docs/reference/index.html | 2 +- docs/reference/summary.xspliner.html | 72 +--- docs/reference/xspline-1.png | Bin 51926 -> 52208 bytes docs/reference/xspline-2.png | Bin 51926 -> 52208 bytes docs/reference/xspline-3.png | Bin 51926 -> 52208 bytes docs/reference/xspline.html | 56 +-- man/summary.xspliner.Rd | 21 - vignettes/automation.Rmd | 2 +- 27 files changed, 658 insertions(+), 726 deletions(-) diff --git a/DESCRIPTION b/DESCRIPTION index 712a72f..2c6bb2f 100644 --- a/DESCRIPTION +++ b/DESCRIPTION @@ -1,6 +1,6 @@ Package: xspliner Title: Assisted Model Building, using Surrogate Black-Box Models to Train Interpretable Spline Based Additive Models -Version: 0.0.2.9005 +Version: 0.0.3 Authors@R: c( person("Krystian", "Igras", email = "krystian8207@gmail.com", role = c("aut", "cre")), person("Przemyslaw", "Biecek", role = c("aut", "ths"))) @@ -20,7 +20,7 @@ Imports: stats, magrittr, purrr, tidyr, - pROC + pROC (>= 1.15.3) Suggests: ALEPlot, factorMerger, diff --git a/NEWS.md b/NEWS.md index 353fcae..e56a594 100644 --- a/NEWS.md +++ b/NEWS.md @@ -1,3 +1,11 @@ +# xspliner 0.0.3 +* Summary method extended with comparison statistics +* Extended methods for auto-extracting model metadata (lhs, response) +* Added comparison plot for factor responses +* Documentation extended with new examples +* Ability to specify glm options +* Added more informative progress messages + # xspliner 0.0.2 * Specify parameters hierarchy * Use of S3 diff --git a/R/methods-xspliner.R b/R/methods-xspliner.R index 013ffd2..9feb481 100644 --- a/R/methods-xspliner.R +++ b/R/methods-xspliner.R @@ -66,14 +66,18 @@ measure_diff_roc <- function(a, b, measure = max) { } fit_roc_diff <- function(surrogate_scores, original_scores, original_labels) { - is_score_surrogate <- all(surrogate_scores <= 1 && surrogate_scores >=0) - is_score_original <- all(original_scores <= 1 && original_scores >=0) + is_score_surrogate <- all(surrogate_scores <= 1) && all(surrogate_scores >=0) + is_score_original <- all(original_scores <= 1) && all(original_scores >=0) if (is_score_original && is_score_surrogate) { roc_surrogate <- pROC::roc(original_labels, surrogate_scores, direction="<") roc_original <- pROC::roc(original_labels, original_scores, direction="<") thresholds <- union(roc_surrogate$thresholds, roc_original$thresholds) - roc_surrogate_on_thresholds <- pROC::coords(roc_surrogate, x = thresholds, input = "threshold", ret = c("se", "1-sp")) - roc_original_on_thresholds <- pROC::coords(roc_original, x = thresholds, input = "threshold", ret = c("se", "1-sp")) + roc_surrogate_on_thresholds <- pROC::coords( + roc_surrogate, x = thresholds, input = "threshold", ret = c("se", "1-sp"), transpose = TRUE + ) + roc_original_on_thresholds <- pROC::coords( + roc_original, x = thresholds, input = "threshold", ret = c("se", "1-sp"), transpose = TRUE + ) list( max = measure_diff_roc(roc_surrogate_on_thresholds, roc_original_on_thresholds), mean = measure_diff_roc(roc_surrogate_on_thresholds, roc_original_on_thresholds, measure = mean) @@ -183,27 +187,27 @@ compare_summary <- function(surrogate, original, surrogate_pred_fun, original_pr #' summary(iris.xs, model = iris.rf, newdata = data) #' #' # Classification model -#' data <- droplevels(iris[51:150, ]) # selecting only two species data -#' iris.rf <- randomForest(Species ~ ., data = data) -#' iris.xs <- xspline(iris.rf) -#' -#' # Comparing summaries requires providing prediction function -#' # Prediction as probability for success -#' prob_rf <- function(object, newdata) predict(object, newdata = newdata, type = "prob")[, 2] -#' prob_xs <- function(object, newdata) predict(object, newdata = newdata, type = "response") -#' summary(iris.xs, model = iris.rf, newdata = data, prediction_funs = list(prob_xs, prob_rf)) -#' # Prediction as final category -#' response_rf <- function(object, newdata) predict(object, newdata = newdata) -#' response_xs <- function(object, newdata) { -#' y_levels <- levels(newdata[[environment(object)$response]]) -#' factor( -#' y_levels[(predict.glm(object, newdata = newdata, type = "link") > 0) + 1], -#' levels = y_levels -#' ) -#' } -#' response_rf(iris.rf, newdata = data) -#' response_xs(iris.xs, newdata = data) -#' summary(iris.xs, model = iris.rf, newdata = data, prediction_funs = list(response_xs, response_rf)) +# data <- droplevels(iris[51:150, ]) # selecting only two species data +# iris.rf <- randomForest(Species ~ ., data = data) +# iris.xs <- xspline(iris.rf) +# +# # Comparing summaries requires providing prediction function +# # Prediction as probability for success +# prob_rf <- function(object, newdata) predict(object, newdata = newdata, type = "prob")[, 2] +# prob_xs <- function(object, newdata) predict(object, newdata = newdata, type = "response") +# summary(iris.xs, model = iris.rf, newdata = data, prediction_funs = list(prob_xs, prob_rf)) +# # Prediction as final category +# response_rf <- function(object, newdata) predict(object, newdata = newdata) +# response_xs <- function(object, newdata) { +# y_levels <- levels(newdata[[environment(object)$response]]) +# factor( +# y_levels[(predict.glm(object, newdata = newdata, type = "link") > 0) + 1], +# levels = y_levels +# ) +# } +# response_rf(iris.rf, newdata = data) +# response_xs(iris.xs, newdata = data) +# summary(iris.xs, model = iris.rf, newdata = data, prediction_funs = list(response_xs, response_rf)) #' #' @export summary.xspliner <- function(object, predictor, ..., model = NULL, newdata = NULL, diff --git a/R/utils-formula-build.R b/R/utils-formula-build.R index 1b14c52..f900a0e 100644 --- a/R/utils-formula-build.R +++ b/R/utils-formula-build.R @@ -204,7 +204,7 @@ get_predictors_classes <- function(data) { try_get <- function(possible) { possible_response <- try(possible, silent = TRUE) if (!("try-error" %in% class(possible_response))) { - if (length(possible_response) == 0 || possible_response == "NULL") { + if (length(possible_response) == 0 || identical(possible_response, "NULL")) { NULL } else { possible_response diff --git a/cran-comments.md b/cran-comments.md index b4b4639..feff630 100644 --- a/cran-comments.md +++ b/cran-comments.md @@ -1,39 +1,42 @@ ## Test environments -* local Ubuntu 18.10, R 3.5.1 -* Rhub Ubuntu Linux 16.04 LTS, R-release, GCC +* local check + Ubuntu 18.10, R 3.6.1 * win-builder + R version 3.5.3 (2019-03-11) + R Under development (unstable) (2019-09-02 r77130) + R version 3.6.1 (2019-07-05) -## R CMD check results +## `R CMD check xspliner_0.0.3.tar.gz --as-cran` results ``` Status: OK ``` -## check_rhub() result +## win-builder result ``` -* checking CRAN incoming feasibility ... NOTE -Maintainer: ‘Krystian Igras ’ - -New submission - -Possibly mis-spelled words in DESCRIPTION: - interpretable (9:41) - Interpretable (3:15) - -* checking examples ... NOTE -Examples with CPU or elapsed time > 5s -user system elapsed -xspline 5.92 0.164 6.217 - -0 errors | 0 warnings | 2 notes +* using log directory 'd:/RCompile/CRANguest/R-oldrelease/xspliner.Rcheck' +* using R version 3.5.3 (2019-03-11) +* using platform: x86_64-w64-mingw32 (64-bit) +... +* DONE +Status: OK ``` -# win-builder result - ``` -* checking CRAN incoming feasibility ... NOTE -Maintainer: 'Krystian Igras ' +* using log directory 'd:/RCompile/CRANguest/R-devel/xspliner.Rcheck' +* using R Under development (unstable) (2019-09-02 r77130) +* using platform: x86_64-w64-mingw32 (64-bit) +... +* DONE +Status: OK +``` -Status: 1 NOTE +``` +* using log directory 'd:/RCompile/CRANguest/R-release/xspliner.Rcheck' +* using R version 3.6.1 (2019-07-05) +* using platform: x86_64-w64-mingw32 (64-bit) +... +* DONE +Status: OK ``` diff --git a/docs/articles/automation.html b/docs/articles/automation.html index dac9949..0afd4d3 100644 --- a/docs/articles/automation.html +++ b/docs/articles/automation.html @@ -88,7 +88,7 @@

Automate your work

Krystian Igras

-

2019-08-31

+

2019-09-05

@@ -106,16 +106,16 @@

Local transition and effect

In previous sections we learned how to specify parameters for the effect and transition of single variable. Let’s name them local parameters.

A quick example you can see below:

-
library(xspliner)
-library(randomForest)
-
-rf_iris <- randomForest(Petal.Width ~  Sepal.Length + Petal.Length + Species, data = iris)
-model_xs <- xspline(Petal.Width ~ 
-  Sepal.Length + 
-  xs(Petal.Length, effect = list(grid.resolution = 100), transition = list(bs = "cr")) + 
-  xf(Species, transition = list(stat = "loglikelihood", value = 4)),
-  model = rf_iris)
-summary(model_xs)
+
library(xspliner)
+library(randomForest)
+
+rf_iris <- randomForest(Petal.Width ~  Sepal.Length + Petal.Length + Species, data = iris)
+model_xs <- xspline(Petal.Width ~ 
+  Sepal.Length + 
+  xs(Petal.Length, effect = list(grid.resolution = 100), transition = list(bs = "cr")) + 
+  xf(Species, transition = list(stat = "loglikelihood", value = 4)),
+  model = rf_iris)
+summary(model_xs)
## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ Sepal.Length + xs(Petal.Length) + 
@@ -123,23 +123,23 @@ 

 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.66812  -0.07307  -0.02386   0.10341   0.45717  
+## -0.67011  -0.06856  -0.02511   0.09891   0.45816  
 ## 
 ## Coefficients:
 ##                       Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)           -1.32737    0.21623  -6.139 7.56e-09 ***
-## Sepal.Length           0.03222    0.03547   0.908    0.365    
-## xs(Petal.Length)       1.85685    0.35456   5.237 5.63e-07 ***
-## xf(Species)versicolor  0.08438    0.17014   0.496    0.621    
-## xf(Species)virginica   0.39232    0.22838   1.718    0.088 .  
+## (Intercept)           -1.24091    0.20204  -6.142 7.44e-09 ***
+## Sepal.Length           0.02889    0.03555   0.813    0.418    
+## xs(Petal.Length)       1.84981    0.34695   5.332 3.65e-07 ***
+## xf(Species)versicolor -0.01324    0.18502  -0.072    0.943    
+## xf(Species)virginica   0.30189    0.24099   1.253    0.212    
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03096157)
+## (Dispersion parameter for gaussian family taken to be 0.03078324)
 ## 
 ##     Null deviance: 86.5699  on 149  degrees of freedom
-## Residual deviance:  4.4894  on 145  degrees of freedom
-## AIC: -88.655
+## Residual deviance:  4.4636  on 145  degrees of freedom
+## AIC: -89.521
 ## 
 ## Number of Fisher Scoring iterations: 2

When the black box model is based on higher amount of variables it can be problematic to specify local parameters for each predictor. Also formula becomes large and hard to read.

@@ -149,15 +149,15 @@

Global transition and effect

Its more common that we use similar configuration for each variable, as it’s simple and allows us to build and experiment faster. To do this you can specify xs_opts and xf_opts of xspline function. Let’s name them global parameters.

Each of global parameters can specify effect and transition, that should be used for xs and xf transformations respectively. Then you just need to use base xs and xf symbols without parameters:

-
model_xs <- xspline(Petal.Width ~ 
-  Sepal.Length + 
-  xs(Petal.Length) + 
-  xf(Species),
-  model = rf_iris, 
-  xs_opts = list(effect = list(grid.resolution = 100), transition = list(bs = "cr")),
-  xf_opts = list(transition = list(stat = "loglikelihood", value = 4))
-)
-summary(model_xs)
+
model_xs <- xspline(Petal.Width ~ 
+  Sepal.Length + 
+  xs(Petal.Length) + 
+  xf(Species),
+  model = rf_iris, 
+  xs_opts = list(effect = list(grid.resolution = 100), transition = list(bs = "cr")),
+  xf_opts = list(transition = list(stat = "loglikelihood", value = 4))
+)
+summary(model_xs)
## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ Sepal.Length + xs(Petal.Length) + 
@@ -165,35 +165,35 @@ 

 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.66812  -0.07307  -0.02386   0.10341   0.45717  
+## -0.67011  -0.06856  -0.02511   0.09891   0.45816  
 ## 
 ## Coefficients:
 ##                       Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)           -1.32737    0.21623  -6.139 7.56e-09 ***
-## Sepal.Length           0.03222    0.03547   0.908    0.365    
-## xs(Petal.Length)       1.85685    0.35456   5.237 5.63e-07 ***
-## xf(Species)versicolor  0.08438    0.17014   0.496    0.621    
-## xf(Species)virginica   0.39232    0.22838   1.718    0.088 .  
+## (Intercept)           -1.24091    0.20204  -6.142 7.44e-09 ***
+## Sepal.Length           0.02889    0.03555   0.813    0.418    
+## xs(Petal.Length)       1.84981    0.34695   5.332 3.65e-07 ***
+## xf(Species)versicolor -0.01324    0.18502  -0.072    0.943    
+## xf(Species)virginica   0.30189    0.24099   1.253    0.212    
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03096157)
+## (Dispersion parameter for gaussian family taken to be 0.03078324)
 ## 
 ##     Null deviance: 86.5699  on 149  degrees of freedom
-## Residual deviance:  4.4894  on 145  degrees of freedom
-## AIC: -88.655
+## Residual deviance:  4.4636  on 145  degrees of freedom
+## AIC: -89.521
 ## 
 ## Number of Fisher Scoring iterations: 2

But still you can specify local parameters that override the global ones.

-
model_xs <- xspline(Petal.Width ~ 
-  xs(Sepal.Length, transition = list(k = 10)) + 
-  xs(Petal.Length) + 
-  xf(Species),
-  model = rf_iris, 
-  xs_opts = list(effect = list(grid.resolution = 100), transition = list(bs = "cr")),
-  xf_opts = list(transition = list(stat = "loglikelihood", value = 4))
-)
-summary(model_xs)
+
model_xs <- xspline(Petal.Width ~ 
+  xs(Sepal.Length, transition = list(k = 10)) + 
+  xs(Petal.Length) + 
+  xf(Species),
+  model = rf_iris, 
+  xs_opts = list(effect = list(grid.resolution = 100), transition = list(bs = "cr")),
+  xf_opts = list(transition = list(stat = "loglikelihood", value = 4))
+)
+summary(model_xs)
## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ xs(Sepal.Length) + xs(Petal.Length) + 
@@ -201,23 +201,23 @@ 

 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.67121  -0.07495  -0.03046   0.09856   0.45944  
+## -0.67309  -0.07436  -0.03025   0.09788   0.45994  
 ## 
 ## Coefficients:
-##                       Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)           -1.46014    0.26089  -5.597 1.06e-07 ***
-## xs(Sepal.Length)       0.31132    0.31424   0.991   0.3235    
-## xs(Petal.Length)       1.80229    0.37818   4.766 4.53e-06 ***
-## xf(Species)versicolor  0.09996    0.17451   0.573   0.5677    
-## xf(Species)virginica   0.41648    0.23573   1.767   0.0794 .  
+##                        Estimate Std. Error t value Pr(>|t|)    
+## (Intercept)           -1.368640   0.266276  -5.140 8.74e-07 ***
+## xs(Sepal.Length)       0.278368   0.331361   0.840    0.402    
+## xs(Petal.Length)       1.808687   0.374137   4.834 3.37e-06 ***
+## xf(Species)versicolor  0.001331   0.192390   0.007    0.994    
+## xf(Species)virginica   0.323242   0.251553   1.285    0.201    
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03092842)
+## (Dispersion parameter for gaussian family taken to be 0.03077369)
 ## 
 ##     Null deviance: 86.5699  on 149  degrees of freedom
-## Residual deviance:  4.4846  on 145  degrees of freedom
-## AIC: -88.816
+## Residual deviance:  4.4622  on 145  degrees of freedom
+## AIC: -89.568
 ## 
 ## Number of Fisher Scoring iterations: 2

In this case last_evaluation variable will be transformed with thin plate regression spline (bs = "tp" is default for mgcv::s) with basis dimension equal to 10. At the same time average_monthly_hours will be transformed with cubic splines.

@@ -227,7 +227,7 @@

Default transition and effect

As you can see in project objects reference, you may find there xs_opts_edfault and xf_opts_default objects. These objects specify default parameters for xs and xf transformations.

- +
xs_opts_default
## $effect
 ## $effect$type
 ## [1] "pdp"
@@ -239,7 +239,7 @@ 

 ## 
 ## $transition$monotonic
 ## [1] "not"
- +
xf_opts_default
## $effect
 ## $effect$type
 ## [1] "ice"
@@ -267,11 +267,11 @@ 

 Transform each formula predictor

 
If you want to transform each predictor of specified formula and not using xs and xf variables you can omit it using consider parameter for xspline function.

 
Possible values are "specials" the default one and "all". For automatic transformation of each variable without specifying xs and xf symbols just set consider = "all" and pass standard formula into xspline function:

-
model_xs <- xspline(Petal.Width ~ Sepal.Length  + Petal.Length + Species,
-  model = rf_iris,
-  consider = "all"
-)
-summary(model_xs)

+
model_xs <- xspline(Petal.Width ~ Sepal.Length  + Petal.Length + Species,
+  model = rf_iris,
+  consider = "all"
+)
+summary(model_xs)

 
## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ xs(Sepal.Length) + xs(Petal.Length) + 
@@ -279,34 +279,34 @@ 

 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.67513  -0.07534  -0.03094   0.09496   0.46835  
+## -0.67517  -0.07722  -0.03076   0.09586   0.46877  
 ## 
 ## Coefficients:
 ##                       Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)            -1.4577     0.2641  -5.520 1.52e-07 ***
-## xs(Sepal.Length)        0.3332     0.3179   1.048   0.2963    
-## xs(Petal.Length)        1.7661     0.3734   4.730 5.27e-06 ***
-## xf(Species)versicolor   0.1181     0.1716   0.688   0.4926    
-## xf(Species)virginica    0.4411     0.2321   1.901   0.0593 .  
+## (Intercept)           -1.36474    0.27156  -5.026 1.46e-06 ***
+## xs(Sepal.Length)       0.31522    0.33594   0.938    0.350    
+## xs(Petal.Length)       1.74404    0.36633   4.761 4.62e-06 ***
+## xf(Species)versicolor  0.03733    0.18736   0.199    0.842    
+## xf(Species)virginica   0.36985    0.24541   1.507    0.134    
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03099558)
+## (Dispersion parameter for gaussian family taken to be 0.03090884)
 ## 
 ##     Null deviance: 86.5699  on 149  degrees of freedom
-## Residual deviance:  4.4944  on 145  degrees of freedom
-## AIC: -88.49
+## Residual deviance:  4.4818  on 145  degrees of freedom
+## AIC: -88.911
 ## 
 ## Number of Fisher Scoring iterations: 2

 
Then each predictor is transformed with xs and xf symbols and use of default parameters or global ones when specified.

 
xs is used for integer and numeric variables - xf for factors.

 
By default xspline function tries to extract the data from model (rf_model) call and xspline’s parent.frame() environment then uses it to determine predictor types. So to be sure that variable types are sourced correctly a good practice here is to add data parameter, storing black box’s training data.

-
model_xs <- xspline(Petal.Width ~ Sepal.Length  + Petal.Length + Species,
-  model = rf_iris,
-  data = iris,
-  consider = "all"
-)
-summary(model_xs)

+
model_xs <- xspline(Petal.Width ~ Sepal.Length  + Petal.Length + Species,
+  model = rf_iris,
+  data = iris,
+  consider = "all"
+)
+summary(model_xs)

 
## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ xs(Sepal.Length) + xs(Petal.Length) + 
@@ -314,23 +314,23 @@ 

 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.67513  -0.07534  -0.03094   0.09496   0.46835  
+## -0.67517  -0.07722  -0.03076   0.09586   0.46877  
 ## 
 ## Coefficients:
 ##                       Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)            -1.4577     0.2641  -5.520 1.52e-07 ***
-## xs(Sepal.Length)        0.3332     0.3179   1.048   0.2963    
-## xs(Petal.Length)        1.7661     0.3734   4.730 5.27e-06 ***
-## xf(Species)versicolor   0.1181     0.1716   0.688   0.4926    
-## xf(Species)virginica    0.4411     0.2321   1.901   0.0593 .  
+## (Intercept)           -1.36474    0.27156  -5.026 1.46e-06 ***
+## xs(Sepal.Length)       0.31522    0.33594   0.938    0.350    
+## xs(Petal.Length)       1.74404    0.36633   4.761 4.62e-06 ***
+## xf(Species)versicolor  0.03733    0.18736   0.199    0.842    
+## xf(Species)virginica   0.36985    0.24541   1.507    0.134    
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03099558)
+## (Dispersion parameter for gaussian family taken to be 0.03090884)
 ## 
 ##     Null deviance: 86.5699  on 149  degrees of freedom
-## Residual deviance:  4.4944  on 145  degrees of freedom
-## AIC: -88.49
+## Residual deviance:  4.4818  on 145  degrees of freedom
+## AIC: -88.911
 ## 
 ## Number of Fisher Scoring iterations: 2

 
@@ -340,13 +340,13 @@ 

 
In some cases you may want to transform only quantitative or qualitative predictors. Looking into default parameters xs_opts_default and xf_opts_default we may find alter parameter for transition.

 
The parameter is used to specify if predictor for which xs or xf was specified needs to be transformed or used as a bare variable in formula. You can specify it in the local or global transition parameter. In this case using the global one is more reasonable.

 
So, in order to transform only continuous variables just set alter = "always" (what is default) for xs_opts and alter = "never" for xf_opts:

-
model_xs <- xspline(Petal.Width ~ Sepal.Length + Petal.Length + Species,
-  model = rf_iris,
-  data = iris,
-  consider = "all",
-  xf_opts = list(transition = list(alter = "never"))
-)
-summary(model_xs)

+
model_xs <- xspline(Petal.Width ~ Sepal.Length + Petal.Length + Species,
+  model = rf_iris,
+  data = iris,
+  consider = "all",
+  xf_opts = list(transition = list(alter = "never"))
+)
+summary(model_xs)

 
## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ xs(Sepal.Length) + xs(Petal.Length) + 
@@ -354,33 +354,33 @@ 

 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.67513  -0.07534  -0.03094   0.09496   0.46835  
+## -0.67517  -0.07722  -0.03076   0.09586   0.46877  
 ## 
 ## Coefficients:
 ##                   Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)        -1.4577     0.2641  -5.520 1.52e-07 ***
-## xs(Sepal.Length)    0.3332     0.3179   1.048   0.2963    
-## xs(Petal.Length)    1.7661     0.3734   4.730 5.27e-06 ***
-## Speciesversicolor   0.1181     0.1716   0.688   0.4926    
-## Speciesvirginica    0.4411     0.2321   1.901   0.0593 .  
+## (Intercept)       -1.36474    0.27156  -5.026 1.46e-06 ***
+## xs(Sepal.Length)   0.31522    0.33594   0.938    0.350    
+## xs(Petal.Length)   1.74404    0.36633   4.761 4.62e-06 ***
+## Speciesversicolor  0.03733    0.18736   0.199    0.842    
+## Speciesvirginica   0.36985    0.24541   1.507    0.134    
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03099558)
+## (Dispersion parameter for gaussian family taken to be 0.03090884)
 ## 
 ##     Null deviance: 86.5699  on 149  degrees of freedom
-## Residual deviance:  4.4944  on 145  degrees of freedom
-## AIC: -88.49
+## Residual deviance:  4.4818  on 145  degrees of freedom
+## AIC: -88.911
 ## 
 ## Number of Fisher Scoring iterations: 2

 
For transformation of factors only:

-
model_xs <- xspline(Petal.Width ~ Sepal.Length + Petal.Length + Species,
-  model = rf_iris,
-  data = iris,
-  consider = "all",
-  xs_opts = list(transition = list(alter = "never"))
-)
-summary(model_xs)

+
model_xs <- xspline(Petal.Width ~ Sepal.Length + Petal.Length + Species,
+  model = rf_iris,
+  data = iris,
+  consider = "all",
+  xs_opts = list(transition = list(alter = "never"))
+)
+summary(model_xs)

 
## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ Sepal.Length + Petal.Length + 
@@ -413,8 +413,8 @@ 

 

 Model based dot formula

 
For many existing models in R we usually can specify “dot formulas”, when used predictors are sourced from provided data. xspliner can also handle the form. Let’s return here for iris random forest model.

-
model_xs <- xspline(Petal.Width ~ ., model = rf_iris)
-summary(model_xs)

+
model_xs <- xspline(Petal.Width ~ ., model = rf_iris)
+summary(model_xs)

 
## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ xs(Sepal.Length) + xs(Petal.Length) + 
@@ -422,23 +422,23 @@ 

 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.67513  -0.07534  -0.03094   0.09496   0.46835  
+## -0.67517  -0.07722  -0.03076   0.09586   0.46877  
 ## 
 ## Coefficients:
 ##                       Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)            -1.4577     0.2641  -5.520 1.52e-07 ***
-## xs(Sepal.Length)        0.3332     0.3179   1.048   0.2963    
-## xs(Petal.Length)        1.7661     0.3734   4.730 5.27e-06 ***
-## xf(Species)versicolor   0.1181     0.1716   0.688   0.4926    
-## xf(Species)virginica    0.4411     0.2321   1.901   0.0593 .  
+## (Intercept)           -1.36474    0.27156  -5.026 1.46e-06 ***
+## xs(Sepal.Length)       0.31522    0.33594   0.938    0.350    
+## xs(Petal.Length)       1.74404    0.36633   4.761 4.62e-06 ***
+## xf(Species)versicolor  0.03733    0.18736   0.199    0.842    
+## xf(Species)virginica   0.36985    0.24541   1.507    0.134    
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03099558)
+## (Dispersion parameter for gaussian family taken to be 0.03090884)
 ## 
 ##     Null deviance: 86.5699  on 149  degrees of freedom
-## Residual deviance:  4.4944  on 145  degrees of freedom
-## AIC: -88.49
+## Residual deviance:  4.4818  on 145  degrees of freedom
+## AIC: -88.911
 ## 
 ## Number of Fisher Scoring iterations: 2

 
Good practice here is to provide data parameter as well to detect predictors classes, and model type (classification or regression).

@@ -450,11 +450,11 @@ 

 
  • xspline provided data parameter, excluding formula response
    • 
 
 
    To assure correct predictors usage, you may also specify predictor names vector in predictors parameter, and data (optional) to assure source of variable classes:
    
-
+
    model_xs <- xspline(Petal.Width ~ ., 
+                    model = rf_iris,
+                    predictors = colnames(iris)[-c(2, 4)],
+                    data = iris)
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ xs(Sepal.Length) + xs(Petal.Length) + 
@@ -462,23 +462,23 @@ 
    
 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.67513  -0.07534  -0.03094   0.09496   0.46835  
+## -0.67517  -0.07722  -0.03076   0.09586   0.46877  
 ## 
 ## Coefficients:
 ##                       Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)            -1.4577     0.2641  -5.520 1.52e-07 ***
-## xs(Sepal.Length)        0.3332     0.3179   1.048   0.2963    
-## xs(Petal.Length)        1.7661     0.3734   4.730 5.27e-06 ***
-## xf(Species)versicolor   0.1181     0.1716   0.688   0.4926    
-## xf(Species)virginica    0.4411     0.2321   1.901   0.0593 .  
+## (Intercept)           -1.36474    0.27156  -5.026 1.46e-06 ***
+## xs(Sepal.Length)       0.31522    0.33594   0.938    0.350    
+## xs(Petal.Length)       1.74404    0.36633   4.761 4.62e-06 ***
+## xf(Species)versicolor  0.03733    0.18736   0.199    0.842    
+## xf(Species)virginica   0.36985    0.24541   1.507    0.134    
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03099558)
+## (Dispersion parameter for gaussian family taken to be 0.03090884)
 ## 
 ##     Null deviance: 86.5699  on 149  degrees of freedom
-## Residual deviance:  4.4944  on 145  degrees of freedom
-## AIC: -88.49
+## Residual deviance:  4.4818  on 145  degrees of freedom
+## AIC: -88.911
 ## 
 ## Number of Fisher Scoring iterations: 2
    
 
    In above examples each predictor is transformed by default. You can exclude needed, by specifying global alter = "never" parameters, or bare.
    
@@ -490,8 +490,8 @@ 
    
 Omit formula
    
 
    As we could see in previous section, using dot formula the predictors are sourced from provided black box. Why cannot we fully extract formula from black box? We can.
    
 
    Let’s use previously built rf_iris model:
    
-
    model_xs <- xspline(rf_iris)
-summary(model_xs)
    
+
    model_xs <- xspline(rf_iris)
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ xs(Sepal.Length) + xs(Petal.Length) + 
@@ -499,23 +499,23 @@ 
    
 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.67513  -0.07534  -0.03094   0.09496   0.46835  
+## -0.67517  -0.07722  -0.03076   0.09586   0.46877  
 ## 
 ## Coefficients:
 ##                       Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)            -1.4577     0.2641  -5.520 1.52e-07 ***
-## xs(Sepal.Length)        0.3332     0.3179   1.048   0.2963    
-## xs(Petal.Length)        1.7661     0.3734   4.730 5.27e-06 ***
-## xf(Species)versicolor   0.1181     0.1716   0.688   0.4926    
-## xf(Species)virginica    0.4411     0.2321   1.901   0.0593 .  
+## (Intercept)           -1.36474    0.27156  -5.026 1.46e-06 ***
+## xs(Sepal.Length)       0.31522    0.33594   0.938    0.350    
+## xs(Petal.Length)       1.74404    0.36633   4.761 4.62e-06 ***
+## xf(Species)versicolor  0.03733    0.18736   0.199    0.842    
+## xf(Species)virginica   0.36985    0.24541   1.507    0.134    
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03099558)
+## (Dispersion parameter for gaussian family taken to be 0.03090884)
 ## 
 ##     Null deviance: 86.5699  on 149  degrees of freedom
-## Residual deviance:  4.4944  on 145  degrees of freedom
-## AIC: -88.49
+## Residual deviance:  4.4818  on 145  degrees of freedom
+## AIC: -88.911
 ## 
 ## Number of Fisher Scoring iterations: 2
    
 
    Works! Can it be simpler? Actually not because of black box based transformation and theory, but we can provide some model based parameters upfront using DALEX’s explainer object (see next section).
    
@@ -525,29 +525,29 @@ 
    
 
  • Building the black box with formula
    • 
 
  • Black box training data stores factors in long form (the factor is one column, not spread into binary wide form)
    • 
 
-
    Most of R based ML models satisfy the above condition. One of exceptions is XGBoost which in actual xspliner version needs to be handled in the more complex xspline call. You can see it in use cases
    
+
    Most of R based ML models satisfy the above condition. One of exceptions is XGBoost which in actual xspliner version needs to be handled in the more complex xspline call. You can see it in use cases
    
 
    
 
    
 Excluding predictors from transformation
    
 
    For this example consider again Boston Housing Data from pdp package, and build simple svm model for predicting chas variable:
    
-
    library(pdp)
-library(e1071)
-data(boston)
-svm_boston <- svm(chas ~ cmedv + rad + lstat, data = boston, probability = TRUE)
-str(boston[, "rad"])
    
+
    library(pdp)
+library(e1071)
+data(boston)
+svm_boston <- svm(chas ~ cmedv + rad + lstat, data = boston, probability = TRUE)
+str(boston[, "rad"])
    
 
    ##  int [1:506] 1 2 2 3 3 3 5 5 5 5 ...
    
-
    unique(boston$rad)
    
+
    unique(boston$rad)
    
 
    ## [1]  1  2  3  5  4  8  6  7 24
    
 
    As we can see rad variable is integer and has only 9 unique values. As a result spline approximation may be misleading, and not possible to perform. We decide here to omit rad variable when performing transformation, nevertheless remaining predictors should be transformed.
    
 
    At first setup model based transformation:
    
-
+
    xs_model <- xspline(svm_boston)
    
 
    ## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots): A term has fewer unique covariate combinations than specified maximum degrees of freedom
    
 
    As we can see, the error was returned due to the form of rad variable.
    
 
    How to exclude rad from transformation? We can use xspline’s bare parameter, responsible for specifying predictor which shouldn’t be transformed.
    
-
+
    xs_model <- xspline(
+  svm_boston,
+  bare = "rad")
+summary(xs_model)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = chas ~ xs(cmedv) + rad + xs(lstat), family = family, 
@@ -559,10 +559,10 @@ 
    
 ## 
 ## Coefficients:
 ##               Estimate Std. Error z value Pr(>|z|)   
-## (Intercept)  1.027e+02  1.203e+02   0.853  0.39357   
-## xs(cmedv)    9.507e+01  3.677e+01   2.585  0.00972 **
-## rad         -8.456e-04  2.149e-02  -0.039  0.96861   
-## xs(lstat)   -1.402e+01  1.149e+02  -0.122  0.90292   
+## (Intercept) 50.9339706 61.1731524   0.833  0.40506   
+## xs(cmedv)   48.2617778 18.6661709   2.586  0.00972 **
+## rad         -0.0008446  0.0214853  -0.039  0.96864   
+## xs(lstat)   -7.1136654 58.3310899  -0.122  0.90294   
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
@@ -583,21 +583,21 @@ 
    
 Integration with DALEX
    
 
    As mentioned in the previous section, xspliner is integrated with DALEX package. The main function from the package explain returns useful black box data (such as bare black model or training data) that can be used by xspline function.
    
 
    Just check below example
    
-
    library(DALEX)
-rf_boston <- randomForest(lstat ~ cmedv + crim + chas, data = boston)
-explainer <- explain(rf_boston, label = "boston")
    
+
    library(DALEX)
+rf_boston <- randomForest(lstat ~ cmedv + crim + chas, data = boston)
+explainer <- explain(rf_boston, label = "boston")
    
 
    ## Preparation of a new explainer is initiated
 ##   -> model label       :  boston 
 ##   -> data              :  506  rows  4  cols ([33mextracted from the model[39m)
 ##   -> target variable   :  not specified! ([31mWARNING[39m)
 ##   -> predict function  :  yhat.randomForest  will be used ([33mdefault[39m)
-##   -> predicted values  :  numerical, min =  5.975977 , mean =  12.65206 , max =  24.93637  
+##   -> predicted values  :  numerical, min =  6.213462 , mean =  12.6528 , max =  24.49117  
 ##   -> residual function :  difference between y and yhat ([33mdefault[39m)
 ## [32mA new explainer has been created![39m
    
-
+
    model <- xspline(
+  explainer
+)
+summary(model)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = lstat ~ xs(cmedv) + xs(crim) + xf(chas), 
@@ -605,22 +605,22 @@ 
    
 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -11.2276   -2.3748   -0.6923    1.6867   21.1417  
+## -11.3904   -2.3979   -0.6708    1.7025   21.1816  
 ## 
 ## Coefficients:
 ##              Estimate Std. Error t value Pr(>|t|)    
-## (Intercept) -15.95068    1.44893 -11.009  < 2e-16 ***
-## xs(cmedv)     1.70141    0.07246  23.480  < 2e-16 ***
-## xs(crim)      0.54515    0.14498   3.760  0.00019 ***
-## xf(chas)1     1.42731    0.69468   2.055  0.04043 *  
+## (Intercept) -16.91139    1.61080 -10.499  < 2e-16 ***
+## xs(cmedv)     1.72018    0.07356  23.386  < 2e-16 ***
+## xs(crim)      0.60945    0.15948   3.821 0.000149 ***
+## xf(chas)1     1.41359    0.69607   2.031 0.042802 *  
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 15.41681)
+## (Dispersion parameter for gaussian family taken to be 15.48482)
 ## 
 ##     Null deviance: 25752.4  on 505  degrees of freedom
-## Residual deviance:  7739.2  on 502  degrees of freedom
-## AIC: 2826.1
+## Residual deviance:  7773.4  on 502  degrees of freedom
+## AIC: 2828.3
 ## 
 ## Number of Fisher Scoring iterations: 2
    
 
    You can provide your own xspline’s parameters that overwrite that sourced from explainer.
    
diff --git a/docs/articles/cases.html b/docs/articles/cases.html
index 28f7361..624af1f 100644
--- a/docs/articles/cases.html
+++ b/docs/articles/cases.html
@@ -88,7 +88,7 @@
       
    Use cases
    
                         
    Krystian Igras
    
             
-            
    2019-08-31
    
+            
    2019-09-05
    
       
       
       
@@ -100,12 +100,12 @@ 
    2019-08-31
    
 
    
 
    
 XGBoost case
    
-
    library(xgboost)
-library(xspliner)
-library(breakDown)
-HR <- HR_data
-
-str(HR_data)
    
+
    library(xgboost)
+library(xspliner)
+library(breakDown)
+HR <- HR_data
+
+str(HR_data)
    
 
    ## 'data.frame':    14999 obs. of  10 variables:
 ##  $ satisfaction_level   : num  0.38 0.8 0.11 0.72 0.37 0.41 0.1 0.92 0.89 0.42 ...
 ##  $ last_evaluation      : num  0.53 0.86 0.88 0.87 0.52 0.5 0.77 0.85 1 0.53 ...
@@ -117,17 +117,17 @@ 
    
 ##  $ promotion_last_5years: int  0 0 0 0 0 0 0 0 0 0 ...
 ##  $ sales                : Factor w/ 10 levels "accounting","hr",..: 8 8 8 8 8 8 8 8 8 8 ...
 ##  $ salary               : Factor w/ 3 levels "high","low","medium": 2 3 3 2 2 2 2 2 2 2 ...
    
-
    model_matrix_train <- model.matrix(left ~ . -1, HR)
-data_train <- xgb.DMatrix(model_matrix_train, label = HR$left)
-param <- list(max_depth = 2, objective = "binary:logistic")
-
-HR_xgb_model <- xgb.train(param, data_train, nrounds = 50)
-model_xs <- xspline(HR_xgb_model, lhs = "left", response = "left", predictors = colnames(HR)[-7],
-                 data = HR, form = "additive", family = "binomial", link = "logit",
-                 bare = c("number_project", "time_spend_company", "Work_accident", "promotion_last_5years"),
-                 xs_opts = list(effect = list(train = model_matrix_train)),
-                 xf_opts = list(transition = list(alter = "never"))) 
-summary(model_xs)
    
+
    model_matrix_train <- model.matrix(left ~ . -1, HR)
+data_train <- xgb.DMatrix(model_matrix_train, label = HR$left)
+param <- list(max_depth = 2, objective = "binary:logistic")
+
+HR_xgb_model <- xgb.train(param, data_train, nrounds = 50)
+model_xs <- xspline(HR_xgb_model, lhs = "left", response = "left", predictors = colnames(HR)[-7],
+                 data = HR, form = "additive", family = "binomial", link = "logit",
+                 bare = c("number_project", "time_spend_company", "Work_accident", "promotion_last_5years"),
+                 xs_opts = list(effect = list(train = model_matrix_train)),
+                 xf_opts = list(transition = list(alter = "never"))) 
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = left ~ xs(satisfaction_level) + xs(last_evaluation) + 
@@ -174,19 +174,19 @@ 
    
 
    
 
    
 Partially constant approximation
    
-
    library(DALEX)
-library(caret)
-library(xspliner)
-data(apartments)
-
-set.seed(123)
-variable <- "construction.year"
-regr_rf <- train(m2.price ~ ., data = apartments, method = "rf", ntree = 100)
-model_xs <- xspline(regr_rf, data = apartments, bare = c("floor", "no.rooms"),
-                    xs_opts = list(transition = list(bs = "ps", fx = FALSE, k = 20, m = -1)))
    
+
    library(DALEX)
+library(caret)
+library(xspliner)
+data(apartments)
+
+set.seed(123)
+variable <- "construction.year"
+regr_rf <- train(m2.price ~ ., data = apartments, method = "rf", ntree = 100)
+model_xs <- xspline(regr_rf, data = apartments, bare = c("floor", "no.rooms"),
+                    xs_opts = list(transition = list(bs = "ps", fx = FALSE, k = 20, m = -1)))
    
 
    ## Warning: Column `orig` joining factors with different levels, coercing to
 ## character vector
    
-
    plot(model_xs, "surface")
    
+
    plot(model_xs, "surface")
    
 
    
 
    
   
    
diff --git a/docs/articles/discrete.html b/docs/articles/discrete.html
index 5891449..0e11903 100644
--- a/docs/articles/discrete.html
+++ b/docs/articles/discrete.html
@@ -88,7 +88,7 @@
       
    Classification and discrete predictors
    
                         
    Krystian Igras
    
             
-            
    2019-08-31
    
+            
    2019-09-05
    
       
       
       
@@ -103,8 +103,8 @@ 
    
 Qualitative predictors
    
 
    As before let’s explain the approach basing on a random forest model. For this case we use HR_data from breakDown package.
    
 
    Let’s load the data
    
-
+
    HR_data <- breakDown::HR_data
+str(HR_data)
    
 
    ## 'data.frame':    14999 obs. of  10 variables:
 ##  $ satisfaction_level   : num  0.38 0.8 0.11 0.72 0.37 0.41 0.1 0.92 0.89 0.42 ...
 ##  $ last_evaluation      : num  0.53 0.86 0.88 0.87 0.52 0.5 0.77 0.85 1 0.53 ...
@@ -117,8 +117,8 @@ 
    
 ##  $ sales                : Factor w/ 10 levels "accounting","hr",..: 8 8 8 8 8 8 8 8 8 8 ...
 ##  $ salary               : Factor w/ 3 levels "high","low","medium": 2 3 3 2 2 2 2 2 2 2 ...
    
 
    and build random forest in which we predict average_montly_hours based on last_evaluation, salary and satisfaction_level predictors.
    
-
    library(randomForest)
-model_rf <- randomForest(average_montly_hours ~ last_evaluation + salary + satisfaction_level, data = HR_data)
    
+
    library(randomForest)
+model_rf <- randomForest(average_montly_hours ~ last_evaluation + salary + satisfaction_level, data = HR_data)
    
 
    We’re going to make some transformation/simplification on salary variable. To do so, we need to transform formula passed into xspline function.
    
 
    Similarly to continuous variable it is enough to use xf symbol on salary variable, i.e. use formula:
    
 
    average_monthly_hours ~ last_evaluation + xf(salary) + satisfaction_level
    
@@ -149,13 +149,13 @@ 
    
 
    In order to customize variable transition, just specified (inherited from above functions) parameters inside transition parameter of xf formula symbol. For example to use “fast-adaptive” method for groups merging with optimal partition at GIC statistics value of 4, we set:
    
 
    xf(salary, transition = list(method = "fast-adaptive", value = 4))
    
 
    In below example, we will transform salary predictor with cutting of GIC statistics at value = 2. As in continuous case we need to use the formula within xspline function:
    
-
+
    library(xspliner)
+model_xs <- xspline(
+  average_montly_hours ~ last_evaluation + xf(salary, transition = list(value = 2)) + satisfaction_level,
+  model = model_rf
+)
+
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = average_montly_hours ~ last_evaluation + 
@@ -183,13 +183,13 @@ 
    
 ## Number of Fisher Scoring iterations: 2
    
 
    Checking out the model summary, we can realize that “low” and “medium” values were merged into single level (generating “lowmedium” level).
    
 
    It can be also found by:
    
-
    summary(model_xs, "salary")
    
+
    summary(model_xs, "salary")
    
 
    ##     orig      pred
 ## 1   high      high
 ## 2    low lowmedium
 ## 3 medium lowmedium
    
 
    The graphical result if fully sourced from factorMerger. It is enough to run:
    
-
    plot(model_xs, "salary")
    
+
    plot(model_xs, "salary")
    
 
    
 
    
 
@@ -199,17 +199,17 @@ 
    
 
    xspliner can work with classification problems as well. As the final GLM model can work only with binary classification, the only limit here is the number of levels for predicted value (equals to 2).
    
 
    Let’s check below example based on SVM algorithm (e1071::svm), and modified iris data.
    
 
    Preparing data (we drop “setosa” level on Species value):
    
-
    iris_data <- droplevels(iris[iris$Species != "setosa", ])
    
+
    iris_data <- droplevels(iris[iris$Species != "setosa", ])
    
 
    Building SVM:
    
-
    library(e1071) 
-library(xspliner)
-model_svm <- svm(Species ~  Sepal.Length + Sepal.Width + Petal.Length + Petal.Width, 
-                 data = iris_data, probability = TRUE)
    
+
    library(e1071) 
+library(xspliner)
+model_svm <- svm(Species ~  Sepal.Length + Sepal.Width + Petal.Length + Petal.Width, 
+                 data = iris_data, probability = TRUE)
    
 
    When the base model response variable is of class factor (or integer with two unique values) then xspliner automatically detects classification problem. To force specific model response distribution you can set family and link parameters. In this case we can use xspliner in standard way.
    
 
    As each predictor is continuous variable, let’s transform it with xs usage on standard parameters, and build the model:
    
-
    model_xs <- xspline(Species ~  xs(Sepal.Length) + xs(Sepal.Width) + xs(Petal.Length) + xs(Petal.Width),
-                    model = model_svm)
-summary(model_xs)
    
+
    model_xs <- xspline(Species ~  xs(Sepal.Length) + xs(Sepal.Width) + xs(Petal.Length) + xs(Petal.Width),
+                    model = model_svm)
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = Species ~ xs(Sepal.Length) + xs(Sepal.Width) + 
@@ -221,11 +221,11 @@ 
    
 ## 
 ## Coefficients:
 ##                  Estimate Std. Error z value Pr(>|z|)   
-## (Intercept)        0.7125     1.3240   0.538  0.59049   
-## xs(Sepal.Length)  11.3979    16.4565   0.693  0.48856   
-## xs(Sepal.Width)    7.6981     4.7353   1.626  0.10402   
-## xs(Petal.Length)   2.9544     1.2031   2.456  0.01406 * 
-## xs(Petal.Width)    3.3565     1.2797   2.623  0.00872 **
+## (Intercept)         1.369      1.742   0.786  0.43189   
+## xs(Sepal.Length)   11.796     17.032   0.693  0.48856   
+## xs(Sepal.Width)     7.967      4.901   1.626  0.10402   
+## xs(Petal.Length)    3.058      1.245   2.456  0.01406 * 
+## xs(Petal.Width)     3.474      1.324   2.623  0.00872 **
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
@@ -237,7 +237,7 @@ 
    
 ## 
 ## Number of Fisher Scoring iterations: 8
    
 
    Simple plot for Petal.Width shows that approximation almost fully covers the PDP.
    
-
    plot(model_xs, "Petal.Width")
    
+
    plot(model_xs, "Petal.Width")
    
 
    
 
   
diff --git a/docs/articles/discrete_files/figure-html/unnamed-chunk-5-1.png b/docs/articles/discrete_files/figure-html/unnamed-chunk-5-1.png
index bf49c8caab256250f265529a2b7a0c15954b0b09..dca4c32e0811d53ce5f763ba8f44dfadb5113daf 100644
GIT binary patch
literal 93825
zcmcG$byQVr7dMJUiXb7w|Th=zs6IQ91s5i)4ds
z_P@8YkNW@jwoWAZ#((c!-wWUR@9i$KFY)B0v4-OBvCQ4U-HwP@+
zyJ^P|Aqj`|JKa=i@9N_7x$^8e@;X=>h@_T3VZ?&l5WaN0JUcwvn(B*XUF=QZn@kJ)
z^odBo-LX)o9tRiqP@4&uGY1%?sT|LYmhVW{UB-
z*lUaDwlOK>U}0ICt@nTrhbx_)o`woSrSSLn$0Fmea^9Lu6z~|&f7sdC`2myId8s{O
zetv$ENdW0>;yP=8f3acP+M3zrGM2lliV7aA8xD?Aeto~q_r$KgzE!y*`{mAb=@_;y
z2?+_JA*h}UZm)v@NukRFTn|}2z4Yk6U$ZJAidS_Ex_Hy+C=u|Tod@Y&7D$^{|Fx856!LNY~@Xw_ri}}
zH+p-is6^REiHN`*$;il5zL{D3`tIF3G+e5+++-vqr`x{BCn!QfLIeZ^CZ|gsk;}`=
zBa;b7@hq&Y`qv*3HF?3ew6r7>$L8O_!NGB{9HYC^n~=M_T~Sf7+M8fG@coXq!{NpV
z1_9HqmK4%kUVq_R7w=P2erGAB#x8&U{Q2WY)U9LV#-wU&{c$_v=
zjEt6=1F&=(Jk_
zJUlzw;B|gf?|z7mfwAbd9|U!YOD!iZA(4@mR_A^wMHkKc>{)en_2Kpm9}iEJa;99(
z(#4;`#@4nr;)f4CcYmS?xF7VVe}irNcXwCs1MR1eA8BZ52TJJHkT5ecGQ>XIooMv&
zDG$|DS9kl}kS$w6yek7d%>2
z|E90Tocc|E^&ZD;U2JiigXz+6*TeR@j6SA^E_)q?ZDHg#a}5H#yu44JJ{1z0>*(m1
zneqDZ_6B^}!Qa0_b9O;NL0F``cw}UWQBhG55&9-3D;pb=aP7x)K8<1l_co*8bGzf%
z1MZR5c%3_|sj0nu8CYKK6!d@*`R+p&uQNwtVq(Uy_8tFEg4-#>EHYJ?{nMC%E-$0bakmnP8JkLobetgtFBFCTGbGHeU&X>aeR>Ratga_DnX7a?X3`^@rHlhs(|B85xMq@d(>?^C2Fdp-702&+626T?L_`yhObc{i}
zCd%U&o=SCXt;4W9^c84`Z3!)@yF+`*RHK5}crx%-1+LI8j5xCoW$KgXE8}
z=pCN@>Y18~&ybgsQyZ`GIDS*Z$;PI{6tlXz8upO&U26DF{jrXk+LzWv3l4`BnXjg>
zFj`Du!NFyfm9Jc;?_fN5@dU?Y4$6FcrZ!jF+|trxrJK;Y^T(n;df2h#<~3`E2|7AD
z4^<*A>TRfY<99?Pl$4Zpb#+x;4@gM9efg5nbJ{eO)qp~jret1TQ}5>PPWAMu-KyK!
zfw{7>vb8-iuVdD0h77(@hW=fcIj=P}Egg8o#F}9Y*+yyoZj+#l2qt8gpup9+i3y9H
zW+^HvTGp`?#N3cv3k{;H{2s^0mIdgw7Ro|?;ckCdUZyK~|1oX;fH^JH?1$~*a(r^)
z%6JRY$y=hPpS3^P7|kt@QotDTfn&hML`n7E)UZA+k_VjNa-bMnti&=8o
z+eFaf%FK)UOHKP=rZ~AVX;#Q|l{|a)Ou*+#D~KC=R`5`ue|N@Zj`o6&h>#GT#?jiq
zV;Y+Fsi~=7zr-m-^z@QNrEgb2n=~{ubasaa@qkg|V11}sn<87GHR)cAQ_V3p1?^-_a(0zh~kB4X0
z?dSazPz39Mg7r;IoIt-zAu$KDxdt!z2gY5&1u+YJ4%wdoukE`j3NU@(65{BaggxX{`fJ!VZ56@DLIeY_O<5*ayE<79cJM1#i#y&3u+9d-Wo*`GpHlPae_YW`88
zDTu!Ks!)fN*RiOyRC#=3XD5tW_*&|aw?8}~q37y*Sy|Zyd;?5^adRmtsfsrwK#cdLhTpKVyZj3bbAhaoEg$qf#R<%bH>zOknawW|VHX!C!wY2I~>M(j6z}
zBMEScC(&vzy{56WPS)el-nfp3BKV=kGROr;ZJL&6F@W~n(r8PDR@#S8g
zALlX<*NtpACHs2aveMJDJ=#=vqQ%9*8Nctx+<2j6%;s#!ZfLj|M8KTW6(Vq<-{^fA
zHB^L^x$P_Dk#OM&VnW
z&YQ0@So7Y&_Vo|&X}s7EYQ34&e2;_0`{XyuvJq^(PTaEdv$Gr$k@sFVWKf^E?k3w4
z%`&UFHVPSBhYg6*i8DRK34FrB5}v>YKc1ooX2ZH##0$s64b^BAufvV#(o%NWX9mAe
zolI}oFbg0P1qTOj=9&YZ*qk00+mZXq5qRFeSz)A{A)_c4{_R`$o2Q3E@Hv+(h0QuwwCi@kh_cJ4vP!xPJgp1Z4hxcggdnFC$-wrI-kuRvT9H;b3X$s)aH
z+1Tj*g}RL^?Y>9xX8p-GF#9d;avM3$G%6v78FsBmC)^|l*}Gcr@#3`P8)Yp
zu=zcAO<$jAVY)0tr-m(lD9edBWpMLcJP<$gXZz_*c9
zvfr8el7#vabe`b!BndUvr00Bbne&Qa3J3_um3BVa)vu8;#3f4Lb^7GZ4gIb-EWP*L
z9gMa8e9g*V2CR|>A%_l|lvhnvRyn?eAJYxGBmkKx!Y3dYGWP%3
zGD6G9IC2$1?4m{lK-O!oqZZKUTyyh#z8)Cgy$$U`z-zbLwnp=bYoES*mmwQpqI-{A
z@Y`d0|6k|_cK?(Yf;TUl%i@(;O?>78Z;5cx~~A+!2%nQKT9z4b8`y
zFlq9O%S-d_{r%;xiHQli$N}N^)zzoTBJaz(`FVN2GP&%`Ch%=Hn~fEaZuO|@==5HO
zI*j!Jwpi<{PDx4O>bV^ifT5;PX{thQ#z3*7nzqA>f`p`Jc-;*pFuXs&dDXjdz7Xls
znR<(+w6c<%-<8G7?fl4kCM=f)C*`KUFU5E%3ymSxp9~pt0y=&GFrXz6c!;A#SINiC
z%<5lXUBZydQ@b%gCm!bef#@#s>FH|H*u1c+s%kA@<;pjtl3!sL40+>z`SNA+%5cNg
z!vm%9CuxoHmzbEC&!0tgb1ZGslYzr?@G`}885$Vy;CQ&ZtL47n6nk8#lzq@1@wgLh
z{F3U!JyHt`iyY(L&ABGI1YYy%n4~0f?Uk#IrGg~Q$~ObXy1KgCHZmWe`>~@&@&cnd
z^lklz(G(yU4cF1ousR{Az6l72wsP
ziHsE;+c)95td9&QO--VqQPYQ!U%p7hlzY0MQh0lNZ-$zpL%-a(6@u9D
z2imJ}UsjvKViFP`xN1R>(FaINf~KXV1+z3=R&BbmwA^;DgBjqNb>eQMkMrSbx}OGF
zr3S5PPB}*O*BNqJ2V2{$Z6@@y;~nyc4_#q`XcX-tUbZMG0e}GR{PQyR*}}9buYBCE
zpr+;w5Csqig5yF{?_3Nui%Lq8k0~dqlO)*0s(|g-R98g;utLS;d-uplxruS^mzAFQ
zN~?)|xM{y|c&f%mMl>WbGd7}&Z#km0m`szjdp?lq9Y?CPY71>%qc|rxEWS{OElX}2
z16wx7|G$DA1Yi{%6Jt5|ML7W=G!comI-{27evm*d1vfP{b&VEqKvmWBWh6o;={W+r
znmKaXp^S(^3D83bfl)7a#Z1sU*xTO&csc*$hUmxCR6|?aT@W44j;v<{GWUcP)OEc^qQ
zaCD=>?VG-tnX5m1Z>_?L03`rCoUc{=`7!-b6;=7}TzA9cd^OJqfuEV%toAy#-2#e>
z-ne!B%MT(huajA_=>h=u5|JwsSFMkHJf;(cR{Hkn=FOYq%uY!+d}qpZBoCJsl$l~I
z<37)byp3g{FnzaLkx@?|!&Ti1YB_c(ubS%zHKZ*IjS$)iCT7oR`Y0s$SDYTTHa%
zn#i6V<^bHco-ETxM)?#PD);iG`FIgU={k)0?sxwFEpKn!D(f2>90XjWKo`x%oFwE^
zV24cQU0mB|HYXLn_|JM=E!KZhfZe3f^(`U6%hMAC$G8}1UuXMW%;h_2$-x2?>7g4P
zEtyAsHf4>EOk)&i@J2NrOfa`6R&+`$E&c#Ox^**h<%XI?uD93ubwxN`!px0->?-`l^|lNKs0DNh4Uh{-zhEDvTp<%xSWSq=@HuTbF16p^en#73z}lvEgjbcFoef=q2qQc$
zuKFy*daA-)aj9`)YN}Kcd9(y$4rtSa>`c|_NxtmnoBL(Qexa9CU%74MVip$`9BH&=
zi>P-3!@|a>&UBQNnl6K0SDGHE&Hrz;O!^Xg$O3H#zCY(A^VoK
z)_EK>W{U}5R>YhfOX@PT{udD{CSV7O1Gxac@&|spIlZ}Pdk8W2Mv+P~)**NQ?#Z5E
z4ZrmyhjYJlIuxkA)#1^R35vUX0&jXT5_A*w3*_?*4e10vmy9mSFtQPo7*M(Tm#JjG
z76pvwYl04yf=fDA?GR8smmy2cz{q%hyrVnAKhaF9U1cj0uf1e{3ZgDkgGl`M5n&Li
z&2jGnAOy`Mq*h#9+}4FJZ;IWWe)qo>Glufn{GFR9HN5s$ND?
z&3T_Qdguzo+_TWU#N+_)U~t)_=PL+C#arqCD%bZzN%=TT9^St%5(8>gHmg9}Z;)%0
zl$5H5Ua6>Dz>YXDP7?IW@er>!s8r{1T*I!N`^C3Ym%3*uM|gfajFc~<$7r}Ml=RO~
zraXM$`Oy|1JWXvAE_iJNc<{=W4T^@RXHN~h^#8e0FL)-#^6KTVn|tE9UvWo<0Z9Xa
zukMTV4nfkvqT&-GC3%KXpwK$Z@Sy=p~QTF?E_ca9}5q~zsw&su@@
zaejFAmCITx2EbMs`AfW6bXeFQWQuBw1jeqeCrLh69SJ7DfE`akLg#VaeI&UMfJLrT
z;~3a*7Y*%qua^4j*AKU70K;&a_L6KLz-~55qN*)B{0aDd-H-eC@830j+yk{PHss{w
zCe=AP7PED3)9dL<>wweNyFg9ML`E?-9NPuTF=6-A*tGZCGoLGdmk%H@9g^>;`oT=Pe+Mt}(=dT`;X{fgZWd
z0_96B_w9u)8J}}-s__g!oQa{n+sMczDEG6lh`6j?Oy#t+whodRfY}88O|Hm?o4#)^
z&k4xLpj>gB8vAhqji#&Y99M`_6izy*lZNM$t$zOexyiEO0lFeE%4pif#l<;~9f1(S
zQ6Muub+7JkkbtZ6D^I-?I%<!sgFWX)~vG
z^2PMEh6Z3hnpHQD4pJ0eLk?)mkIMkK`&mneIsWgT-->|AEOCDLr$dPLlondDu#zv9GCTEmUKo*?nv6p
z$l-L;MozdKRk=seQuvfSF!&L-r$U!*Zp1bLNX+A4?WbIY>z+ZhWZs)gt!jIJQeNx+
zigLQ6yVXaDWp)=h=xyQr9!-6)xbs8BNiElqu@4>BTU}RKgHt>CTSjBl}
zd%Rdo|Hnrr?V2K=>D-_h9gy8gNoXgym!PPoC*aScsC3FLqU~>bGpiif`KZiXvpS2L
z1F=Ha#MiJsIX@?VW-re0pmYe2fWT&|B4068VlxWB&4{a+Ywhryu$Y`&Xmm9G`E}py
ztT%XYNuK*1&?G=>Z}KLQMI{vSuJiplhshyMsR#}g0h3l%_!c}kJV7Tg-T+#_n(2Wt
zF|LfwE)hn?7}*gO)JFDv6h7=|a{|<-M1Hq@_|JF;TxuHsjv>Hs57^IzxOq^|*hqVG3lplWlkuxul3+eleXH<$EheAS{il4p9Yy8kTLYRg!aCD`+H3p06SV*}$Oe
zmw=tT`@1>Ny|uM<(-N%$5O`*0X4mpQx%Y{0!LUI8qt){Tc93~~ztehz?YJ>)KA83r
ztd_4V`Zqmtf(Y47JdbBw4xYLsr@&eqsNajs2m4r9Sh(v8V|15|sp&CbIEx0MYlT<7
z$j=H1=+3Hy7P$MvBO_Big^Udihkztz?-wC`u^i@~JGBiA4DghoH+M`=YsEJc
z5)oa&3?ji-I^{>b`@bwer7v=|78y4752m7!5Ls2#=$c=Xx8X;*+1s;i)n_Q$+7{W*
zf5}n5G#k2G2tylq4UF~-OK7E#{VkpgIxaCeUCP#D~#mX_ntea@SIx=_HtBn*NC=-o54qgj#D7}jFglV
z$aL~@b*^&2zt5;{j(m?8p|vY2DsC1(;UMC3W@y8_Q&+#S^AOW79A410if*NMg0G23
zOx#sAI8p78$>lH$)k+Q3yi9icXK(Kt=(5m}6}Xd{I-}{$AG5vztI>UT-naEWuE)`)
z+n-TJr@bKecLGVc6iW#%v-&R2vkB$p`}kb88PrSs&gnZwMx0=O!AQv&<&NQkF5?J%
zNK#TVS`#q$17;o1(|uzw#J_*1tkIddg|39voArO=W9qw@n1CuTVTi!cdSO^4LW6gY
zl$V2w$^;7Dakv!*wk086KN%?ek0pY^M~wpKujV$?A*2ZS+z+xMA{5IzJAy3+N=@ZM
zIlzDgT%4~_o}<=VF7XH({7K+DAgO3-Xyiz@F8Y5;+zu;i6Er@v>3#qVb+woDXgG1h
z)zx)#s&WUIOwF8vu)ve>v2!
zLKXsg$0Up;BGtY|NoVIH5Dwugp=o@Mj#kyuva+_`T37&2HRi@Go$4=s>-xkgNbOK)Cw)u^5SOx{~Jcp;{+~2VSpnW78ceA
zH`Cspm6zw}jCj2Pi2yh+=mb%k(YT^Y%F1!9?eQGuO*=o%!Px>q2H}~2WeTd2F*1)K
zQ@a@OlcuJ%Tk!?BtQ!2
z=;-X><>ErlqW1%~wPu-Fiif~3xum%;#|OHm{{^ms6Pc*t`pN?=UIuVKMGKAp`$CVDKesO=F+mPy0tR^nSsRi-NS7Qh
zh6%BCfugx-uJv!w=Pf4WBf^L{%)(=02ugw0BV&?s#0gX5SlUO32T_#B|
z!SDn1zt?-s7tP05+6>K*F}M
zjm>or_uQ~{_P;-NgYP!cd?Ui)0$!g0eN&uLf0Q3=^TpPXjX^$M-b6l^-rnAV3UpA|
z;kA1tGKO{{4^O|=?XL+mgQ`fh+wbl_mH@Dp3ZbELWtM-htrAaz$>8Kaudy+1$e%!-0fam!ZAfy#>uu#*VYdUmf8N9Y`wBntzxV&g
zsi*&6;#3Oq5FbLleG>~StcD47O^PaVYrHrWh|7Zq59H+JpkPLyR;uM|zypGR@2T>i
z9A{7Y;QW`?vJeAvd_F|z>g0U^=*0}B_At83m2m>aOJEy?hhuYdb4R>@M9eS<8K4V*
zf0;F93_a?mEIiR{`#JE~KmbuuQ?rkeVv2ED>lb;1{f*yEDRU?{MlJ6(D*zB!)99!u
zBU96u=;-^R-@ku1+p71ztSRU^8C5t|%!!-4M!F*SFLky{jzdnq1K|$nMnDRocc`eU
z8k?E|-L(YCsOKvupWT9AXFcFgjrWG208IDw_5GU^gZl+C0S$Qr2fEfd^cZ*u;Fy|W
zhN>pLeDWFkU?{eMzCJ7P8Q8sCkh&?rxRWd?;iUcV!e%`M-3!nyd?K~q_{7BF=0qtl
z6fl86OqG85vZJeun9~BHHAc`TqFkQ%#m2`wOjB^1^}oHehera&V;VDT6F{fDj%!8m
z{J^!R+th@P3T#45Ozep1nh3lBQzA!1P$pL`UrA066f?cIrayLfeE$vD30oIcR9tOU
z&c!bCx+OElbTv0M#nOnGnIYKZ*5+ocESMs?5CYm5$<58lG3otw3I;o%co)2&7G|Bg
zM}BvSIH2*1O_KZpuq)H92XY6*IQDjTKZb;8!zFA77V}-n}>{=6~UpQM@L5$m6e4-_VxK27pb?2pz9!B3dDdoA!bGTU>+V6
z3=#@Iuxvd+Q>Tph_y#HnZcqv|GuwG#1A~Q#$C9nVgkqHqu6v@C5!j*^7Z+vxqkvOU
zDcDWA$4iVcN;9BPV2l14FJ6Rg>asOiZqS16dGOHHvzqnK!EU+Vt>
z&d_Z2oqGhm^aaaf2B10NZqLjIpHN+P#V|()Q9KCZww@d=w|wQ{;o;!$7XXG#0kTm>2yU=(a|2h{0wja)JY8lUDg4N6Crx394CGYXq`yeDHXV=F|uyF-kOUiOR87*)`H=vQ*raNJK=$?CdOn
z>jZXFIZsc13M1H?-HIyC-}Qrm2flOwOjc8pmSzM{8d6$&dwZ?l)QhF!J*ANl8d}UA95GDgzq4vC{k;)Ma>WYylpsw!eZ%Cmxq9__8{$
zcBZS9^({fgiIs(_(;tPv+DQp|)G~@RLRu*@m1*{l@
zR;8M#;pSHD?|+Bl*Wh4DUS6=e8<>eJy3f-=$^wglZYgy?DXpx`5hUFWxm5^UWw(0#
zK>IJy%LKW)Zhu!mIQU$}Hf6J*YaC^Cc1l5xX+=-$=Xb5aB@3VDkssMy)
zfH07#GIDZ0B|u3e2C{*bX#exw9X?13lo$m@M&g#n0I7i~!$my^UV%vr5SVOfu>iUD
z-rfv)=tM73p-w?$h4Ft671b7Mo0Ic!fB!3oIZnN)i6Pkl;I&9nK`=3Fot>%t@^tFI
zsy~jF0@q9y^_jy;FjwXe3R>Dgu8+1(PNd+K02W0g`pnH=&Ivl}v$38R+w|j%0nO(tgNhbF7Jch4#T3izdsz?z}h+?F0Lc3
zucHHo0~!{VDcc3?n7GQSv^2k$$q>KTofiT5degwt64Z}*n39k^j3In81f=2TPmxnH
zCA#Rr*Pj(=Hs>0z!RQ6icP3kz+FgR&bJsW2RHbyXKz)M8_BK(O5J3pJuy&3kL^(jg
z@(YLV90nuhbblF^MFugvSTwR<32-lPQ
zbULQgE(wvG;VeZNcW7fpMMcaFCkF8MGQM$rRohA8$p0m`iCDOsMN^8Zs$vTOZ0i;D
zhJ=KmQ9!0(Q@cQ?zS9cO7>;)K{x%I8+vk_n5U}%K93)0s@%}J--M+jJKHp>n1wX%5
zeCfky_D>?OVDQl4vMh?TjfW7pA8*T|-VYStaE}f6@B!?k;?hrnf%%X{(z9M
z+N6gFxSQxlK%264kaZGm2Vu0FR;Uu-Vg?IrTP8-v^gpGgrL|6fTCL8&^j|xJ#WFDl
zL^poy!4?-h8-?!Q1PauVVyp|K_IDUYmnF_?+XR5hsw42{a%-TCT
zQu+#j7eKsqJp;l>uvkMDuc`vlWcQ}Z%>T~+ymgcT4&h)s0`E8qrEZKm_jI=Uf_pgQ
z@cRU5Ln%ENN<)q^g2qVtoCYdJ4whq;14MfDzhE7+Tf+J>MoUcuLX&Fd@KHFqipWLh
z=x@aXmS`Z#GJJh?EM$UPTOPsmU*f)B57Dw!dCu*~y<~d$cd7lJJbgNluj_NkS^C-Q
zmF}YO`%bk{$D1Gfgs$N!kJq`eM+!oGhO0y`)y2Rs9Cpp%;r8~nV|{%56oeEIX}ay&
zof2Y4gwDD6$iCVW053YO4vr{DXFD^sZBJ-uXu!LH+>3l7|3JnC2@IOnrMVO_l;wP|
zJ3ut@3!k6YGm(YHK4b(*{4ZYA@&hiOFuper4@g1T`?{|Ku+pT0>2=ZD`UVZ(4*(3=
z5OQqPT*l|mIFs1#y|C2P)LI=qAQ=UomLk&>gh=9)w}-PMq2y;K3+(GLroMyE(*h3T*Aedp{t9u-fhIt8=ljZHyHr
zZQnTo4JP{4AO$i#J$-q{69$IhIGsHJRNzx(Ixyjg{xaBuEJdFT{?DwO9D@fiz|B@8
zU?#x!Gk`u0z=qd#R}a|b7e>wQo}M^<{MN^)7#JX94uFIL#)G-O{s4UaQ|T}Kpc;T}
zB=ZikIsl_RPksk%&Z2Yvh0HV?Mv~W2+`bw_IN^Dlny!K!-??lriaMm2gv|zr3??U&
ztbw-3L@hxWnl6k7vYE@#GRkgEU;9dsL&*d>89X60S9a?TWL@4vl$B=331BD)O``48
zNuJ+X#3@q&gvv0#1^J{XzCd9zNzutt^SDHnfO#i$yCvf{-5n4^;ak`#oj%uVVxjY_9~)dxc#n&V+l0;m#6Ut)5?2&}UMs6#{85eL`u94eWj$XK82RS6)-j2>
zPR`CiJobk0KjQ&SF@tU?%tqC%geI@hYSSDS*W>B`1F3F6g>)$u)AYf
z4Po4NBhp5lMr=p-0h*V=ga;=CI?Z!QNx-$AP0j%zfok{@#LkHlWAtSiIk|#SD+My0w)GB#hAQq!)hp(P!;3O=1!T|H|+3p9mfUJgaHCr9@LALQ#f8F4DibC!w9PhgJ
zKH{q6p9421E_y^%7my}}*b{J+%Hm?gcHrelK}P->8|%0}xD3(>N5Adg71@>IC)VgG
zz>hIZc7w
zP=hQ6#0#7YfsaT55IhEZ2N5_EmH~aX#JCHNha8?AT8aX#kbV7n970Z#hPWiLfyfC7
z395qos$XDkGH_T7ztAS)WioMhuMuTKK|xv8p{Q%`>4}y~_rtcav2ikoLnjc8#Uty=
zOiN?IfTuF-8U!izqn(}N>>p@0@QA>bc2iOsgh^>LTL;?z_n8T7(HF3NVeJ>8C`1?@Ttg%l
z5h#v{c@;(`0JS*?sy8z;^H%8~D&xQBq&N&S6vI1=WH^34M=OerUJ=upJy6h~yO|h+W{=1dY|;K5{Lxl(aN{Xo1WlY?>r;P?Ck4>Qhh)y~~U2w^ZDDuMAL4kKV{|K5k$5277@Zg&&%Azw3WS9HH_JiccU^6whU?0VkjS
z&fq+F(Api_^$}>!_q#B6AU+0%d{RXPIW6DMZ*2)6Y-WcQY*7$yjv)}V3ALD&MMFtx
z0Ez-69STna?~B3-Jj9R}mX?{H2OkEm1FR%6H=yCz23A(nz%NAGAu9#@T0Z02(+h(G
zVr+eYBu4YpuaAAMGueJZ-Z;)5l*0aKm@@~!clY-Z{zJ{TxQJay7`5IYf@BW_o2B7+
z0ljP39#6#hY%wjq3UKYzT-+dpiv$XLKE5a~L+x
z{rf3wUOGBt4hou@nnw(YXHBQ@a%+~ly(IR#T)5b8oak3%P
z1yIxjRX5jC0bUwdz7$Jf#-E;^HeMbPv-1atHUyW&G)RJ4J}H@cXK%4h^C1?lsp~#GrNV`6lowa$h?X{l;lJi
zqz0m!*S+AZlMKNnp!u2jh*#2PW;CTmAd@ZULv8^?8x6U2>1i7Z)+>+2S3pKjg8PXwOC7Qccfuw<`
zt{)b%r!MZh
zQxw`cRNKK?K{$;3{s#vKaKMu7k_s9?D_NNjFgBT!NBsPr0OirSDFA{ba#}v15PyX2
z38&Vk?|cjky9RQQ@a>!6^@YRH87vv7v4GQfTdy8n^w5Gn>A;3JMCyH2Z~v`OaQyPRC*T$gkf{2al@hiQlibw&V4o0Pk>fEGnn9
zbW;=cLFkn
z5ey9z6BCdS5ZSKxH}@@}J%L>VHr_Hki)-N8fQ=<3+hEcpfag@LO$gJ(!QnZYe-lET
z1eb#;2Be-daMJ*`*f}}L%FBlVRfi_}*=w(bfXMU)WV8#!Db=g(;rs`j)nS4&kPrk1
z1n?WWEh3l;hch6+ltJGH^BB~|_JmS6&F5CGIDv->gI`y74kRnYWtjg!
zq@&#FpQBEIFb)1VP*en%fPw%!1&-hF($oKeNo8keC-W89(cs>xWk`NLWU)0t{s2`M
z>{om69HA0U&(7fRB&?1A96cNyWP9{TxO%7!fG(3(6*exe0}K=RE=1BDmL7Pgpx_mv
z*8SE}!vvKcniU~iC`^^BRS?YsJFOiD0Ry6Z&%mI9qDQ3w5dylG@(WxNk_*5l(OVk&?R0AmH_;lRKEK#D=9j;pAGodJGyI$&!A}?IzZa(r3=X!M
zuG$0h4FbB5=z}IA|BdHo@b>!Rq8Ksjy7<3p-DFDYLu^qlo7udg=%f5ceor7~veU~S
z2PqfRLIxTd(6&u^IWhs_Q!RkauH@i)vOjYJJN@>?@N1MDjzDk?2pkd;+&=3ykr
zap2E)2F!a)#txjLQNVgrcnGk6xII+7Gmy@*olpfZrLu0`tqf9vnAL*XII`HgwfDPCh&8hpJmGybBYE5v=_3Y
z2fymy-wOcK&3fP^6Po3d7dfyX1tDWns9dIIGmAiU+IN8}PL6m}~
zCv-A@OI}`n3;GpEau8+kfJ%qk)u+0$1@jgBVrbbw4q{E$J0Ba2kB`GP-Ud1cDrc?T
zq67$v>FLIxFN18u4!|(fP0S*>rd&JTIQPLeR&qm-05Ei*_ZefK?Hkcrfk4BzUx2SN
z3tR!t6NAhFR|_|kz-oxh{pQ+dS)J40PX|rd=jVzun9y0cW4MA_8_4^
zz!p&O0~ipC>k2pB>Ip}1|Ir=h^VPJpzQx7S;-SOi)*3VZ-PX7sd?mZ8)c2oZ0$*15
zGw
z7XsHRubq)Bxu8_tr%LM$dEbIg?;J}@=>|PDX3MTu7$5(Y78ar!oBBR1F1wXv$xo}D
zdHkI2T<4{^#q)IEw^b0?>FJqsa9*=CWX$r8p!y|r*_gK}LG_(Ln|F9VcsM(}!@IBN
zoRg0_C6B+nq*FndU;MkysM?Trf+iOgZj!st-Z9#2=0NsYk{~5Y(
zD=JJ)7~K=UF-H`=GEq@8R#CfDe0|Wd=Ht>F3(<=qjMF=t(f*%H
z+R}}$W1?rqTHOnEl?fIt$|h-IIzpEl3bq5y@KhBZK1hnyQE$;9q^1p`XFR7BaQG76
zR
zysfUf(Cb1@BThKZv_#<8aQ>)PD$gtXp2_N$7wMCkW38dxd9%e~@O@F=!mra!LqAD}
zcP>w^sxw7~gnB4ZMenIdfMI-w@ehL(E)|=op3BS_ifP`oo-3}jgu~<3MkiTin-=;5
zeS~qUY|ki`fKdNHv2)=oW4&Cx{sleqK%MUdzsuOsgB5$xZM>K~X<2z1
zdTPHkHMI||0)ZAoox(wKc!9>t4kvFOW2-8hh%7k7eZv;@yZQ6BX`AA`BTQLwTKwqM
zzZl*!eR^_#m;Th1XRikl)mX-DeKTbyi~h2wse-uFe+>X_Mw?;V&|3OZE9wh^s|u;*
z{XU(#H0H{4cX=nf@62A^GLk&wJ+@9{_{UEQ%7U?Rbk!QTI4?2ev7D)XKYQkb_kgwQ
zmQ{XO`9W&L$BGrq)@PGg9(X>qjp5k#^+jth-fYvLc!W|eOF!XbFXc)z#(Q~U)vNF{
zQr>l6g&_u4)G9dG+h&CpPijl&+;-cQY`EoD~=0~UMHi+*@2atV>oszWpmn^
zu+WtO;s?TB9|{orsm30ku+mJg_2StpQ%~Q@=fD13r0t`dxD_Q}@^uQI#>J7n83=B@
z$CYW?c{Ne2_uQVYRF4^N`|n>xvz`7Q{kw0I3b*1}dbpW~RK6^o4<8hV;cCnU+BMky
z8CvZiv~cGh9zW`SRW{eQZ6a$PPV=g#Hd6V;6-tAcclUaC(kSlRYj*$-{>DtF%_kt>-$2=-yliV|6
z$m1GVrIo62CUud`7pBay#dMT9tBr;>Z4oK2naKb0q$p56^^j(0o=&-
zy?=J?MKc7Fi`oJQp0_y4ny9M6KPn=>kGk_(FlQRAmQsqkW)?o*b1{rtE-N6N5&slI
zL}yg;^d&QQx@i&lz7QRwhl)nUGXV#e(u`EnEg@<5YPzQ(W?vtbBDXyD>+eRRi2O$C
zV?L_l@VuzMx%we-<SC0TcshhviDYZg+yT73VZ$Y
z%WQtkDbdqQwmt2?fzfjpvoqZX1qH6R@Ut>Q$lcYPMhsQ8r;%Z<2TL+K4W0SRY!k6<
zXc}_QlfxFw-ybQM8TaBFKk{=e`KqGQekGWn^{!FE)Xa?1!ph+L8y9kRWdg&+?i@PN
z_Vjn}+{ivr@7Fna_{WjW%DcKpW!4_!+I*)l`G8!XpU6(m*Us2
zACKy)3*#ti@@^vi@VWVqP2)ohTQ0w(bFh6ELp}_8WC7(OTlS`!-Ja=7=4;QtDn4%Z
zosw^uByL*QPW5?Qy^|oc@|SR!)<&6}pH7e>Pn#{oxmVVW956uO|JtC4wO2
zly8XG-)BVM6ayo|w?M=-^$c86aDKBzATJ$p#aJd&EyNd~&Vn6oVuP+u6`kH_j{Dmv&%hCx>+AgU=llL~@E36OFwZ#Qpk5ygF6WAQtqNsKI3FjT%ts!dBna(Pl{8<--z>%Tf@s02gJaY9Ww_I<+J
z`qTzv71aT=kEO-C79s-S-NM2q)in%KQU%Kk
zd+AD4nu*_Ps5rIHR*ZZsCV$0@Y<`d8q)W@s31plYM;qtc(4?i$icnFkW_Xq^lXJAT
zXS}SvBgFG(?t4IVdQ*m0u`ATTP_VIE>W7&;h680C#5L!V-9a4w*NnRCH6Dn^4o
zbC|ZHSAnJ@fC#q~2rDDUf+7JoOIBOeOYj`J32Sb!t+9A^I`5-8CpIX1^PcDvywaqxIgW
z*O!8hNoOJ+#@AK?`ZU;^=A#_l>pP)zTMXS(>ak(YMdaj3t{YBd6i;%cF0`d4k#|}-
zdd`L&O`_0IQf!lw!fI9!>A>qZhfIjDP2!=FYMUO{=O~=8
z`DuIQEbEA}Ny3tE(CdU}n!92+-*m_~b?Z;)5PDC9Auv6zh}_xPZkt(H*_U7Ik3&mg
zo_r~kJ{f&U!BmS
zCzyRMY0WHqe7&A)@BH@Hd!1`p=6W|lDSGgpb2I>D$#j2TbGoL{AK%y&e*Kz$K*f}2
z{YZ#=dm=uYZXQK{_;r}D*p_xyQ7T1+Xeu+YTbA^?8N)?;PqaEywA3}tMXqdHnV)HQ
zIEjOR7Rtor|Hax{Kvfy8>!K4>FaQOlOQ&>8NQbn9Al)D#-Q6fkBi#+sCEX?6-QC^Y
z=bO0py8E8J&l>C8edlmE{xN}n&i{+|d*gYZXTA70_!1|$%7B=Sm%wPMKh=c4l*Pu3
z`7(%8VQ7YVVDXb%d}v2UTVIH)Bcm(aYvTS%OPf$3HEL#myY<$XVVNa^L6Pt7FUUUc
zFhBiGf60M6I656yJtoBAS1&v*LQS`4qAmE?fd-S&Xf)$Xu4Ky7@*^IEe%L{FR(xQ|_h%6uc*6{ICjS1Tj?9j8
zf;kjo-N?u@0E0y^@3G`O6Cq@M6_j21#xjL1#FmqkJ=-2>LZ
zLx$Wjud@UDn?#2^=RXsbZ7|v)D-rC?UVf-c8Ng6dP9L~6EZ;@`AT>G~r}%}8bZ2}B
zsebx~X`)6e!Z%VT69Qfb4rS13Pcd!gFyTv+0_*rlZ}iN-$$p3
z`SANg#T#bltO864dZan=RYN14;gqk@`6S@;Gb3NT2j@jUmYEFtrJ}2h@|Tzxl$UFV
zPYDeizvrjo7qTYj#$7ncr6m)2!56B!GMzCObG8o59!a$&-p$=b0;At)PlS$A1eQZ&!?>m~yyth~mwzl@zbxW~Coz~@ez&nhx6@-1
z5<4|b8#wi2x$HyeY5MCA{AjPJpTKpaAz?p&LsAF$ZND1PQLUZSiC(t3Iifd#I!du=
zR@PsH3fA5Po_2TJaUVs)2{G*K8bX(r>ejtqL0GEhLsdA*}6CggwT_rMD!PMPJP
zr{x*h+`@9e9eda#c#Z75%A<7))QQ27n~O6PXQz!2R6Oar>8Z1<+PyTMT>D`%h8~aw
z(JCAVrS34x3*ussp8=yNz)4eQzQEVCgTYQiLvvt1ow^7qQbTxzEPw=v83MZ9y?@2#yo5t9Hx{sO7z>!5sA6k*Rux^b0?=h?RQ+G(-GpX;8{Dr
zInx{pxhnIU=ITJI#`$w(HKRN~!6NOf{S
zvlkZV=YPT7^Z;7|JSqgT{EH}G)s-C6Ew4aBKpDNG0H5_ZdR{YH}qaK@5tTU$UNLa7XTMKWq?GXPG=mn_A()JK8>!C3@J
zJkXp0l?EMHQVl4l(WXXreUcwbYWs(rI1|XhQG2~lmP5H0rJ#mOFSvYDeabmELj^s1
zmwSDVoh4vuUcxytVq<4?wdHctw)BLg=oS8?l(e&o&B#8|AC0onSdQvlGmZ;ojX56y
z%LWTdkFRo%9x2F$xd12)CDY{wTX0THObj$wL|Z#`D+dn-WK?DcV8r>$kBEqX%Z4U?
zKjj=ac`{;T@&)Q9U&cA9A$8gtzyvcc-{_|8(cvBW9S4tV_SPAq>3!%n^bE@3J8
zKdTNth!Ry}$ylty;b|d-K-k=29(aKOfu692+h#@K?<8va)x?=q`$c`u*HE}L`DI6#
z|6qyt9;9s4ElTQ)#|VY}yst+Vt1LeLIImre7_LO3rK9T6Tj7B~+P^r5!9g0RexzU2
zg@gcN1Os5K7KUFVUr#|Ef>IlBS-5HUqI{_@O^9T_i!T-s34CRYrrOb0&vrwp)_JPW*
zE1gc}MY2@gXMFz_uf&PMb|$wHOQm*m2T3R3uHi(yfA6tHHUsA#2RJ|WSbut(8y1S6
z#;=_C9Ub0|CBH$P!VLbN`n*LbKUpy$@veA;>Rs5&i-A|}PXLA#p>?*~h8<8B{8{+z
zUE#XeEqfqEzU(*wAQFEx>B(oUQC>|^C*A^=irreewVjBsLk^0akwf
z^wii(KLX#D>JWK@S5g>HO{rU=N_~osicVqexQX~Z0X6)?By{VN;GLw=T_G}@qSO`E#
zN>fu4WamXG?5J3nnL!gi>OsjlK(!z@t<1q#h+YN=3F@@?Vx${Xso?XLbtQqXXI_Ti
zR%kM{M+=&>~nq)D`|TZh)(cx$r(yl;a-LbVT~6|Kc-s`U#QK4K8QkAv)2ZHZMg^3~kix-&2RY(u47+Vqj3D
zsTOd8&h4%uf#%f34xPc&MG8SdLC;3INQh8pQqt=a(v_L3SCHid&=_$xT(-t
zAPvim?OhX#Q*C;0%2zW_D%!G&j>^g_Fbc`@e!g8
zapmyj`OC!=m2Aoa07=Ji|I?tWQ^8asx2>7@o8V(+wTP2h-eau1ojp3Lo0d(C
zqo&tHQxAf~6k#qaWVxdD!GND_&09`gY?y6sC8!Q;L1&!Ca90(C87#O~
z(Wg8`#&)=i9!l^8vZTofLpQ#naxs`|3~^^0%>%xG1%94xfr(_%{8GHymc|^{!tc;5
zIcx~MC(=_~(8umg&U|W33u)K@T?e44O@J65)xMf<3Jye
z##f08aB&pi7humDdd9ny{X8P_BkT6JFk{m*+=?5_+C%9!y$$yQYiC
zNQs&(Vu(0e?0XpGr1;OZ`8r3})}4dW4QQ8t6P=6n$cXQCeJp&C$Njc)S8A4G;K7d}
zBGrP-WBhab>-760v{_u6!a<6}iP3!8mS9{IXi?44pi4!jA8=lZlY$bnYQgl>l#x<|
zcUWX3Ra#M|y(kmzh_aI`KJRH~l4M$$iJ$kVcA^a;#JmbXee?5aQ33J~woU@At*rrn
zB3)fRfI`8{B?T3gRV^T?a$>?g0k<(TD_CMpcWCXPhIE@!$HQg{0?B`UZ@>QM7f;S-
z02|jPQWL$z&_B&@%$MvU>CTfy#(kDQ^yd?34de{u`i~JjKnFiHWH}r76)`cm_0Qc-hir#hBoR&Flw9)dY318k*(gBJJi%KFzAX{D#tUhhv%PC)nLf;lUu
z-vwCWz>p^+7?b4_pi(gOBoSp8i?OmVjzdGtupQzS&>C7$9`L4buEWLlt^&{P63+a`1n_m
zGQ=GVcw7LzfE9jqbtPlAq<%dfLA&dy%XSiDKg{NC9QpnqCa&G~-h=?l#{56LfUC#D
zP6<1#egtV}STL{1%oGto@|eel)Pw+6Wq@o1bocSTdNu2e*4x*|A~AGQw6U1Q!{zLW
zXioZn7ZvQ-UsIfKL8DJjN2efaGEvlW(g+)J$SW24)cfNIQqh`|^xlXdK*&o2
z5D<~_VBh~-Cv^h>nPu9%sp&b7v6B2{41kbb8EITi0wRIEnxo$1@B@2i7>yS`XrBkK
zqG9Vk_F(oz&^9lprPDhmdhW%;K_BItJPxl?1}1K|4zA@t{}+eMMbeLel*cy63L``4Yr)HPHtHM1%o0
zx%kn`z`OH6FuiIx|M>b{)*hY!IO=?Rl!z5f=X&s5U#$&{(!EB2A3%H2($Rt5vONO3
z>9;rk7{M)ZyNza0gI0GwXaPWlo!Q#zq@?)_o|_F&-lm1t-<}kJ#+=Jy+#0~t1bQeu
z`I7B9x#J@sny31pTYF;zHMpv(>iXKbehSWg+8O|l^qHyv;AWV6ot7N{w>DD?uv!2>
zB}6%-#(09E$^0VsbYl_lcQ8u~&A{SM&6qQ{`LBrF;K^K1ax`ej$zKFdnO5%r_}SgQ
z7mxFoPWpxUjXg{_NO6F!5(s1MpFM)mrOg?u0dqHGwv>;Eys@ahzsn34^dKFNsXLZs
zuds~}&$Wl^YA8Vf;T^f*5etjWdAcGw`&HBT?V;0~oy?2y0BcHUK9X5=>z7hcuv?3_`x(`La#G-S^X(b!eh$g~H>PJ61)kb-C=`H{uG)pUL0=|P
zll^kAa6af4s^@^6Z4ihajm$R$)U5xAAPN<`fjbI&HOQXVTmSu?Wu*V}Ed0|{$GFcS
zBXz_bX+U_(ep7c@;G;f4@DQcAUua>D#mGPS@)bUkhGz9OC@FB>Ah!GcBb^bU`^4e_
zBrX6@roc`&@J;B*$@i-XAng(WFBS0DB^0g#$hwd56Sc|7jr6JsMlgy2eWp#dgRfuV
zP?c&OOATUnVMF!TMT0<8`R`Bp|MnvE{oixF^1Ypn?`TRGWx;;N*Sc9=4KsSo;-`Mj
z29O40q5D6tpgy3xBOid1lf}Pmox2!xx0&l(-Ek5Yh%v2ci9~>OfBARL)sCc|R}Wx5
zS=u!${DHANfe1+OpFSDDc1@J+0G|vT6jNn_&Ga>wH)f!XPTg8vMrO%o4N|1=A6cJ2
z3-qX9jnDuUgDuJo3`|VNgSSoy?1(|N|Kj-icU*tq)`KaZ*5LxA)YE2y&59MQ=VThE
zU=|?sTd>%Po=Wlgtakln^>CrX1mc6S>3K6AZn5Saf~okMy^U3E$5&9F@08Ti7*@kY
z7^_t$hNdIU7$vr+^;!>Qp*^X0f-CYUz4*M<
z=p_6LDPv%Awvy{0-fehBEp={oF7kTJEIk3faBk9F>CBsWo0zcGq^aP#Cot_jTdSP
zhi0otTn6-6$MiYJGV_D4on2B|4A?dAOk;n!UaWIQ*KA!S^W6A8^vDSBpRBsdw?;_L
zUW<-YHyx&ZInZ19=Jcp<{q!B`z>M1HjNK!iyE)QgPd>X6r9jT*#i8Tf*SW%k!(__o
z8i5)0115Q=M=IUV)ohdlW;o*b5oOw1eMcInZt3F%qK>eljK;{^4j-&FI6)Wj+1Ixx
zbaRzh)WrtioUXb~%k)Svt=8BaFJAQGaqK9lTc%NfxDnp}vpB?T^;Ml_AH;{KHG;xNi&9ln=QrZxowkZDHu4$*&k4@ko&lnq+8)Ps?zQ>4p?qVue3hRZX1isH&P@BkTmGe6
zZN{Y;ZI0`NYw_ih1rvtl<&y=q#&Ol}?3c(|YxWLOefF+pHd&8L>p!LO7vpU!sFx5{
zNeakXd{}q8-BH9K4)%6iqLcP&>~k4VEYr$7_pk9BT?!oUyLo40oIi(MaWVQTcIJ>!beUEXAex5YH
zxg1AD{WNj}f43s!)c3KTqw`at{vj5R#q12nqT$SpuE}jc?NRCcj@z%!8J?p@kcywc
zQFMRZ+j!GaUB?4}FIO#Vau@g=biWDDTNGv5;>*v*rxM=sn72f;Y%B?*^IWSNO5bE!
z+bJB*RzJA8;g6+2sxQr9yztmdC%X$g?02#yajKG?Su8~8Mv~NMGT7MpaN{P&==LqR
zsjkbQ3)e8$;7;>Lk)_;QyeuZ1n<&w38nmerC*lIFo
zgwVA`e%8Gj6Ei1y0pb@(vzA^D`SY#4gN->ZhKSe^PXlbKO$|QYww-;`xGsu`cglsd
zQ^Eog9)q@t)0@3@ySXRh@+CD|Ggx&@dhg@SdIVBCs@=%TXhd;x`rcxWiK!=;a+}tl
z7R6nb8F1a=S;P&q=xA?Prn**!MQo0upPoj_pmzy2J6=Aq8bTyCpCs0s0Y-D-Y>2`m
zhGnki*fP~L_#u4N&i9+cXC6l{OUeixME*Fe{=kYKjawpiyKG7hd;NlmrLIq3F;%q7
z`q$jVRm9EVJD--gH(qHM6?z5z&73aC86QT5(@7b`YnW1twb#|HsJuXnwhg*riksNT
zw;zssvCtHbNuf$MI9!bb|cB)uk!*TMQ%6vw+E%C{>4RiHr_K%%GDUxMV
zfwz?Vg_1quVi0;=*l^6Tmfaji-s)>xB)|MFz^Bzxb-Q5}%Y8FJvegzBwwnibnCvHu
z`G~22H2y94Zn4?K^V@PnvWpDY%;!JgSzM=8?H2l@B#1SBEU
zyy$RV@6O0^r9L~C0dp)CSZHt{T6wYp0iT%$4WTa8VJ2{!Frka4zYrh&UsPzj@I=NG8Bqdae|WSIUl3s2;q6v(Emzw!)#iC~-#nwza-#4XK`Jyqcvz^toG}tjs?e>NTEF
zE%i&dqiY#F#bpfpfl6<%$irtJH7PoMU4G6Tw7Ws5y!DBGAayPA5^KuB+ANTg>MTj7
zPf3zz0e_kfnq336A0odS)YAIN*lS6K`sDZoVdW}ivs|~X4jKnBQu$&+K
zP31sSJma%*A@p^B*^{zn##}A7@XqE`^hnbBb2L0Ns0Rn$9=tNcrjieEyxEVx8}>!O
z*(nI8q>RuM3Grbxvz*1@RQXzXi7%(byxJ5C$~6+Bsr6{6Gn3GRitkO(kV`oS^LD{V
z|Dk#?xgbuOt?^7Krrz3U+sRWIl;MMCndZDEiuj1MW%vgNxe4+7&^ldM7&^*z$5z4I
zgrR`v7jZrl+!sOJv8bR4nx8s7S~x(CD1Yo>5F>a5a841=j848l5oKNK?IbMCowVu$
zOH=LyI~G2Q-|cOH#c9*CD9cu3U9J}cWQeKg+6yU9a5LG$@eOwHd9;SRP6Kgp=?vLnSNCc<4xj4Gc#2V8Ml?M_mDNOvkMi3
zJ{A>t-vT3}0oNay-%O%xQvo;9{!-GU04vNjv&48c_QR*pXd>=`k_*&hpNQnkXQWSY
zv)7Q-)(NgdlANj#%xH*?!hHELI=fs@a5?RfYnSj%i-=s>*%*!qB*ViHOs~($_0q9z
zpZX7dVeuDg^US-VFWz2?sNID?pxSg*Ktxx<|uoIZTPO(b-4S^yj=uPO2fwjC}TVBuese&5Tl9p8I
z)gOD$R9$W|uTV3k^wyXN2j1S^(q9AJL@^}IU6CvJ!WFQ)
zCI^sC?rf*NZY@oEz!7-MFcW
z(hBG!sF@4uW)&;!udlKOKR@YhX2kuNCMKHop_w0Nd03{noXf~8c7D42w2N+>u0DwU
z9VRWxZAWD74oybZ#WQ#om%+=GzBV@gyo<&;#?jo^j(muwAuK-JAKhmpKrpS7>xp(6
zB{@8liLKhm4@G}YSeBYxeN3$NQ#`}wFyYSHb`QHIO5;VvD#FnpKT?}HDaG)J^j^kn
zvZuI%4}eqnkbmhpf^@eHR`wsQtz@T@=sOIomk;w3j_Y$E^DHn2xrW&}yn1&dfv-Pl
zRetkSK3cV(-kHt4dT_0`UHz0%GED|SP6Z3t>^{Ti@N!CA*i1}Bcy2n5?%BBs(iu@{
z{unuYZU;O&EVk@>HEA4@z9v~kPqkYpt#%$o7yqDyh%
ztbQAg@-G*|v-jHXRl?l1T{js=c2EvT0|SHo0#5SOB+$wPt7
zaYa=ka31#Bo;^%;RWhvRo*Br0bCdmB{}^#|Yb2g*$Szu_i~F(buHWMLQs4*Yy(zRn
zKkvdXYV%DB@l|B#EUIXaqt-O^6Wo<p$g@u$?^rc8F;clT89qpn0}(rsmq$q7@w1
zA8Gs*T1b)$?j$z5zh%2twl8JDol?;~%CB?Ht)7}nz~Q0BP9*g9)(t52`+hS?C6gcb
zWP8muc5+pUZ=r$AH1-c+Kx?*;gJH3^_h@fE!752)0DJeu$z{J9Zg0GDx$CEJIrQgX
z&ne4EWo8@Im^_9A{qXnk0-3Xox3tb0o4n--zsu`6N9kwg^y&{6=hfs{(xnkl-$qc&
zVR(?0m1s<7t4R8Tycexg+bKm!#vca{;h#(Xh2#wrSLxigI!{~Fe!aP=sx{enbE(y{
z)0--rn~rB^8cQhXq)P91kEwrk!!_+{u(eM1_;<)gJn2Zy#Zr8$R(ARcI~bE%q?iG~
zH0!-?aE-tX=;=Ri{
zVl*gJG&+`>-a1rzHN7iuj`I~%OR0nCMrHr`olY}dUsVCi#nK(iLgj%_njzM&z#U;X
zUFU_qL7XC27=zn{}cms%qf(YcQnjLzp`dBKpd-fs
zhiHeB8zr~!cz3)GM&0M0>WHRV;cHPG+Gh~!%H&>WV;j4SK|Qjyql#Aq9vn`p2K?l(_=%*PM8
zZH*3j>d2JE7^1H?o(HrtgDFn^Quh*LE5Wq(j_WFI&E{_2fS_B9j_B>1nVNHHr?#9d
zwnxBHfwYyTdP#@AYO1Ix3Hy9a^_c5I<#060@V6g&94EXzc_H7v4WF5>nado`-A>e8
zI)ZKS@EZfY=AP?}t6?#rcjn>Kj;=A9wiySBFA=-=lO2N-#a$D{Mj>`cwE09&aRvNK
zv9Rl>R`E^IP@SfVtLo|YlXGNs!}aqdnN(B~#1-B}(6Zo;Rh%KsX8_qh5n^eXrCau3
zKyIW+O$r51e?+biGAa3klP8Oy2AuV)x+;c163!o}cY^u>Teo5rbblaFss1uepGFzP
zfVLldmWkL70H5)ES&TH8y1N8WI{AeZ+HF*lf4QS-l$e
zRMjC5B&gGA*}P)o+^+OKiT-gw^F#Uh#LL5NUl_O?orBR{8#6c=Vp$CtJh9ii@mUfq
z@$m&DwsoY9LHc^<)@aOnrt>n1sAhF6@Q|Q|sMUuMcOzL=$({n{UZvD*A*Edz
zbPJm``TX*$=6t7-NBQjQbT-k0!J_%KyKco1*3sGktI_Y=XM#@&-1a
zWtZjYcG28yY9P_jO|8}xWj*gmH%mkw{ip`hC=R-p70R}iZT&72L4R#6NNfC(rBT$+
zpVQA(q=BLl{GJ=9fM)+5XI-rBc1|UI!
zsy)J%U1I&TkK0avr{?S@$Qp3F3Y;B#qsb|M8Kz;N9XD4-_!`$gJPb4jaq;lT>LIUh
zd+Nj1-6<(iT$c7)amw88ZifebK>6R!j$Pz-|EjmnPTG;>wVa#eU%;6Y{t{L`L|)6g`n(J>5{v-mI{gV*Y`KC&jB
zXNe-+saQD){_SxClf?aJ-LxaA_@nr(^?QhWm=rmXkfDR-Czc~Yf-q6M2-*%(
zEJ0`CCgX3r#P2|muZA2C#jAAmkDfgwwF7`RUOawmCI>Np2MB+`>t+`ol&S&8I#+g$
zW+QwcWd@CMiHO4sp|R1?Ds^<;-gzwg`%L`5Du>1(O(Fs8C8Qns%TgtKu+(ha!Dj}4
z5-}1U4FTephdGFlV$A!};e1g;Pxf!1$^ocUzVh-4NZ(;W=6`?|PtZ3@M+wxFR6wd1lyCe1
zUn!{NI4IQo{@wo$=Zh@C)6C$;J7v4^}U
znrMHEGye@N<-Y<|{+)Lu8uozWB(ui)b5t2raHK>pK)P``eK-r|Nj^>^j`}QFccB-i^srg@4U~OgOk3zq(oK#
zAl>F_A-ugnMu%-yZ_B_>kt3a$aA?pYgHEnTXgw;2?$sPIwH
z-$LBS0C`7ozNI`CCNKh;wJ=IkL
zW1g?DySG;kzAw<>43k!pfLduMYt?p$XLHZO7E?T-G?|
z4q)SIq3fX6?a~hwwT5m8gIU=C>^#3pgijcA~r3%9Mw=l(7!t8GhHLZVMq1!>s)4&E
zu*DDvkOI)T_GNT<`xDw)}jT)XDz?3jp5y
zzubec{q-v~wW-VmH7h^=^xi|Lt*sr1xCNleVIa@%{sCB(EtsDQ@P{Cv+Xt`}Nn*4E
zsQ&uqrZ2Gh(n?Ajj+p&3Gx1MTxhMf&;VUSRinS?caL4Dpd<>rDKNn>~-`^V@{!VHnyHf
zOKi31X)U8R0bB#p-<`nuxxPd$Q9r?ae43!3y`cf=_yL5`<9^ZYyN{lVO_4J8i#Ny5
z6gUAB+>QOt1NOPSKTok~8U`_3Pq{vdSoIm9{q{d;~^
z{C5&E6f1UbAVJIk@9pYZ=Z%fPI1E6`3+Ce%aSLOtp|RQYY@!0oI@LGRL%Sm{QJ;8`
z`SPVJsDjn&?1*Td4ptvn)ywzj%m2-ZxH0`qiAk+EuXX@pj6HB9*n%SgHrPt9&6mhz
zU2o5|K@$A(8H9ConotUT
z5Jo5#oBsLSJJ3{Q$8Z8^kj&kHVKX>CtDh!!Ajzy?tU91)00)TX@Ayi>BHmo%D9xbH
z4_)uj(H{4T2t(-6|1tRF7&R-l+8e%a5dq;+tQm7eK}=%;q`_+Y9<6ClO2>0P0a6#H
zrHMCrgx*N?EsAAO^B!cwI7KuZKb7lPLjz=*e7fc<=uX{oE5CLCT|
z#p-U{_HTG<|8!J!RxuQ!Z(;=mR@SbD8ZPkss|2AbnAFE#+9d`n-CjWpU3s(V#dw*cnfC`@`HfY?h<^qX|IP(KS~dg$++>?YnDk#S9MI^)(0c#jECGyf5p|beU=|3gi-Ge|
zL$0o-wqsxc!;sUTub?%E080R58LsYO=YV`22LFPAFclOP*FUYuD#iu0)H+E(mLc#W
z_p9*#qr(kzuYnO}T&u0^sJ^AK!N09G6LaVJ3GOM=rU#l;zmyA1&lb*1+=0?xIwCYg
zB=Zkc?yWG5%IAEQf>n75)%Qkvch$XHQ}L^*Rn=l)$)aKblk7{^&jNI*1hb+u=@;H0
z3B&=LWp@S=hcjbj$w`#7%$UPCYx&Ib@4PKWtvK`kGo<8r>H*|cq>fZSJw5$A>rd%y
z;LF3X_8iSi2RaNCP(ZN;1LLC<1>S6mjjqm#SZkh#SCtE&owg@?oJNrKlx{s3n1y
zE_%)KvhjlYY{RW@EORM_hJ7ht?x7~bDtu)=LaAjYSRlNRRRUdJ>WYe)(~q~mk<;2~v)%0{JG%f^
zCpV2lje2J28|>bRGgEMxc6HV2{M|qy%H4#_%{H?1f+Zsj>5U_+E7QTvT}~h&w_kR(
z2FNrF^gw|q573JV0(}#muq;At06>U49so8eSqnOLz@>cvWN0U7J-z@F>W;JYuP`R#
zAcfZ(3JTgpmqP_2^!Cb6^75+pMD<=|J0^f)j?fMt>JC1K7SEl(sTn#dsEW_HDC{5I
znS^1y7%HGg1?jlBeHn~|ufO|X)pRw8H8M>Jt4;<)61$zg0p%rVRo;lJtKUU(
z&c~cWCjh3<1R35!cKa!!^w7svxA0awm&=l_^Z53X0sxm1&sFWuyd`A&cIPQNXP3tn
zz?{d3HW_2-Ee9pLvT7yUjM!sJc6dYK?T(^SanLhW!)X%m_S-=sXGuX
zew-4R&)$s$Pz%;;9iVub{n@@V2(YqxBwmVN#?}K+3Y*JmY3YhjWAKk3UnQsCS=RBB
zzpPlXr69+~zwQpz<8D5t^6zuK^pTMidc=LZVckhjjEw{UPK8CE&98Il0;+_X0PyF+
zFOm56VsFXq$Y!fhV{0~{cn74clzzaPe8hVj+-t9F^!8$_Y9-E2d=0Www3yw#4w(OR
z*saU@4N*2q+QdvV9qehdRx#QfuQ_L6o)qf+FW`~v
zk8|udPnU)O@@2jbJl`q2cE@0$5Q1q1o5k9kuZf=thuO|s7IjW7_u3N!0w3{13($S^
z&V>>l(m)3s4Mjo1<8
zp#)NA+guma$K>NwY=o}|kWowt;@
z2tcB9lNTd0(}8)(p9zz;Z0N=|K^zgecedQKHLm`1Zgnh{u}I59D6H1QMok*!`r}}O
zA8|;A-d~rh4JV6oHPhkf@a^HN=8
zn~@T?-+W4y*-?z6wP5ASsis^x3N`~<2{sO`JoeoFBb3&P&%WN+%tt#*!h!-UL>xz>
ze_z{`SM}77WN`5baC5luRH1WjRc1v
z(YW1e>eg-D@#Qy0D)H#-EC~r6dig}g#{S-)$xc0a_=elE3Wo1z9tM_bCeL1_d$0d2
zTNslB#&NOf)ClQe8zC7K`a{q|?c`L1n$qDSV&h_~7C_^aBCno+@B=3E}~Ah5%$8qd|~d*(B^wVC2#>*VC*@r*_%`gpXt
zPK{7e^YRdPpSWD)XkkO44i~M9j-bMgYH?ccBxLeohUr+bma8h&bg*Q84L)>95j;D~N?Ba68zaLFD{uW{Arg
z!S|aFA=lk7k5=12EmwB#bof(rT*3*~o;H80v7cs^_o+TaRywDGsDGYsg;eRR%K<44@uPK9NUx6;hgV>Gk%-s3SlbPXFUVQT|-eSJ0%#f_6jE>vp*Bs5v$+5y0rZ^+32ECBv
zCvfBYnz&!5S*Dcp^_oH`_2?52Ow
ztGmMArODoxXL?odPzyaKkB?6yAK2mu;)qFI9LZzVZC4>bu-SfUJ<+A~L~idrB+HD>
z!TKd*#QWQM(}}9^!LtX%$atU1Y^~L4<4*-_NmL1)5~sQwc=u7~O+JN(<~fcy!l~xP
zE6nM^L(Cv?ufVDa(-wl#lT#TnGe3D9mNt^(snn-Ig3GB+jJuw@w*By7d^_%drV0N@
z@t575mFG*p=+KdU;ZS5<$i6;9m;J%X_Q{r0AmNW_fuiCj@`Xr8@5|f|_yqwHQl7M*
zcWz&~I4pcRA$*u5-g7MECSPZQ&WQ4@wj~zMKHK$y$Am^#zDD?56|Gy;zE(J*Zwh|=
zFZJ0$yV*CF2pD!|wT4nR9>QQ)W{pK)F?nWEMFtJUYX+Q8_47DedJdZ>4g+X`nBpm4
zf4UW}!rvH@w5QX>kVhDhw`U9`emX@d@O&2)C4yTiyI_(jG=w3H(h2>e{$q1ecG6BW
z&CAIo(XX@Z&C6P2p--!;S|>?3J_3`C!jp|&FZe`#Hx4a3hqp^dxQ>q0@f(}N7?_7v
zUw|*MkCqN=IwTjaoyHIDXJQ_4dpNrj;`%_W7Kc)avZka;QBDeP(|Lo{&9yDr$(0P08@cTGs$x@KI+r
z;eZ<(Rq`g6eouQ8L;1Y(FKhK(KmKtJ*-|$!N#%rSu&hOA;h2U_AGMTkyuKzK(uc>%
z#yjMA9IJ(#+@8H1UseUAzO07Lev-;`ArWeT4ffVWep8`}X$SVeFWOz<9%i
z=IPkr&-foK3IZQ_9CGW@Q{b<9d8(2mSb&Z1W=c!AI3d1i^B@Qa>$CbM3ol
zH>D{mQ@#ZUYZ-t0*1RNG#7jsdMn?Ir%=l6{K62~*;1i4EBY&(cWFLV=QrDEPt_ey9
zds$d*HU?Vs2JE_~$)BG47Ghg}K~PgnI^?}@D3FiQKb)LRliN#{F1q`pt9SP$LpCgg
z$Db^XZgeTAr2inH0DQU`gczKjkg(L;O$r3&_WV0xeP!K|6zD}QM>oHZ7&_mlxG}JO
z(Uw-|6b$F5-^obCFd}dKfmb^B!$A%aqHSI-^}$XFiof$}*ZKZJOEZ&2%O8`9x=uMU
zBz=WNcDA+4Ox0;*`NZiw`%H@B(n7RlV+kW8Fp}tUtK~{g$o2yM$en>PQWGZ4$Q%Cp
ze!GVJYVwN$XPBMgn
zi{i3VNwXorM!0UKTM?tPTXpheMwADZ9?g7tzqTnqMMOBdAj;x%^FDr!v9eNc7U5_Y
zVsDeQz3n3|#3KIGJ0kvH=3wWGp!!Z+W8fK(Q;(XSb25>vvkaE$oaH}x(ury9+EaX$r8
z&Ia^R%Syl#{9UEPp|-n1dYqwZbs~>nGtu}}LH(*yeh~_8AQQZ{Fi6QJ^0_iHVy9Fa
zlUkhg7a^S$v3W6jjT!GVn=v;FG|dd%xu==i|;+`MuDC
z&J1GNTCYx#xuZ}rk~(}b>DL^0fm#onjXoa}K`oYk$^jD&*{Km_>SrQAMYRBuYk(EE
zriQap)%3otZv*7yS60%}{lMEm^#Z1INk!Gw-`_tlPzbvBZf|b^({^@@z6)+U
zS!hQDtr}2JS1D8m6v8`zqj!@7s5u10#>VbT!OaJJXVkM1P~4IVn|1#j7U0stLndNT
zct|<=&hdd4mAtd8dPV~C%Df1Vn05l?9B*U0$Y8%x!lBlKXFht&B)AMz1;j|eDBP`V
z``%QqJfV}Xkj<5A46-j7#=Cv1tEKG1CCdjLxYKcNDSV-{OdPM*y&(7=8B_AMy5F-eFKqcJiv0!lJK
zRxBRPKmvp_fUFAOw^p{MCAKs(+XG4zhAxw1i&v-+!`Hs)27K*+u3RBc83P?XwL}_d
zNdYx=T?2y-Hv^z300>T!9G+QBjgN!I#J(o}j=%@XM1F6;O@RUuczu7d5ZT42Orump
zOeuO1^64q;4C{>>LH%qZgoV+WTF^l>oyM3j@aO_mOS-8u5s*m|@_w02qSlt%2fdQ$
z(k5!p*GQ9*SqC~jsGWnNi_?mWGam+J$L(~%jyirUDFe(2a7)&w@u3+MVpN@@A2(NM
z3`Qc>xBM(0XMPxQM3A4W%$EPdk;rai`1y14Wx|DzXSQ&DMux(tPwyWukl{Xbefy+R
zA{+qL)C4c68vN^ifJFHzuOQ5KP%UZsVak?K5L$
zAjM54nShOh!)!b*C8q!=0Kt9bPbeWZ2j;7}KOe(EzGsdyngfji*7(Sw50`wnQ&(Lud)l>nojThkEGeESN0L
zbF9!sXeH1s1nMoI6r!z+gNH{oD*}o~wizxobwqGUql0h0gIt5bP6lQgU@sNy+C=pF+Q5V`IZ;`+=CPUMCQTIU;hqehcK|!2E+2
zN|*#Bkz}>@0pLh~+?`=K3G`M$`>y|sMWA2-Xs9Kg!$HiR#IfbP;e$Y4gJrzEu+X?W
z{pXJskUJ620W{*vfZGQMx&f^hpkxD-aO6rZfwm#jX55QGz;F%_5)6U53==T!d%!1u
zc!NH72~Ne!2ia9K){@7F*uKzuR;6(7GFsSfw8*7rqK
z-p2z&L(hn><5*6SnW^A-v%eOLw*W@4M0DynGMGoboQcus$WhegLjp`chA
zeutrtaAD&X>xD_DcwKO0WU9?5d`BkknK!F0b-7EE%VZsfcZS7>Ls(rtE7=Bb8ybS5
zXtXg$ftmRe9e-Ut4T&=lh5tc8K3?AM-(HK)M5cXc`lWXOrh?MTQ+Jd&E@2WpcJ}5s
zs{L7Tu6`$>CUA|M?3;s#V_pC)E18!}yE_G{n>tR7*Q$O=96l`H9Jp`caE>E|W7&H$Z?dybjy|Va2e}6SY_L#^>f7h-rzlhi(5HJwh40}lJFigph
z+yZX7{?7B~&tqc`ko-$aU9sxtmEkhV#~qxN
zRYn7IZD2PL?R=1lDQkX+zL(N8Zmy763SCN{;ED$~bygpz@wT(thdHdyQeuyr97N}{
z|IH>i7$m{+#|BpQ8d_S
zGyzjBSbn3(T3js3DpzWD6!tBSI2V0_;Q&L~yJXa=1J@fL8cE652y(f)-<8pnd+yv4
zI=*ozFz(&!@7ucJnQM1DrxhzE&z(
z#MX@{vNBQOPKx6;*gCNc10}i6Bum)-&Cs5LeZsEd^k}7_`WGRASJ8VaG;XLp6nk$7
z4g$8fy0RZko*`>Gc0J%OYGccYRP`s8F`9%hH7%>GKXoW9Zb#&n*el*blJG&{2w9nW
zzZ1I=ju$bKuEO%_Va};W87(aumgme#AI~tm3&P@FUf4SsKS6a3V+86i)IvSBkKf!s
zn3t?C>o>y_(ZUBbWHN>c@!p+TTGwA#V|P{+Y@SQvXcaA8g+2Z0AjTw{A;n91hjdN8
zC7zDs4vL*R{upJAkK5yDJi7L*V1GKP+)U_)$35mL?0_W~Z9HZZ_Ij?NpPaQkIr@^C
ze2W&}gL6O3es<=M!tj7Lz#O(Qws*Dct(()~nF?2BpJ05?ESy?aXyw%bsC9_M8kG;9E-Hv59P)0TcXSe&QiIs!JCu+G(Yt
zEKN=3l@1J)law@!As%#byGi8a7PrwI6m{+LU2F`xe?RitKKxd|
z50xhpX9dd@UMTJ*Qxku!Ah^8vBfH+_U32p^ys|Q#1oE*k`F0+ZaCurlvbD99mEAy$
z+>8?FR%|tEfBxr>Q?a3nUgsvMD%(dWb_>xlkhSsPAjh~2#f<*?8l9y>o4ih*EqvmQ
zX#URWs+rk)!}+6g##-%)CZi7?JRm&tu=9Q78Vzl!9+u%=9{Zq&=Nk`GYNL?T8)>ZNwe=(|@l}k`;y=WF;&tl?L)*#0qQE!l)?yqy|GHBMh~~
z61>*?#z;@k+Sax)ngPx2+%K%uo`F>c-{!+UWlO(GckI~V6O0>Y
z+Y?BWhpAvdpSwt(vyCSQ1`-MupysHl6Qvu!v&HM=-p1HKffNRup<42|FrZhuxD$Fg
zCJ5go0|0rL(rX#O1@8zS-y_1}IUjnC)Br>Kw4)LZ`}gnXXqHiLgU{i7zJ!CLW7HM`
zg#Iy0OHG{zUi(;g&pgSKJnO{c!qs)OrOFWkuX%qO_yCt;VyAch6VAZ#uf;QnXn=Eo
zb+BQ7?-7=4GvXXa^Se=elC!o@BpgOX(e3+SY}sfy4-1r-xV-ik^sLV@^FnC|c}S%A
zjpOA%VQ3cRXnl|27_Z&7zARsXBTI*WN`Wc*iQ_1;xJA?Hc#bws>-X>9b7$#XxX>s8
zpEliS3$@FvV&ddTtoHH8
z|6v8%g9o2PMV+M_gH!5hT|bndEp%>nzIHunW%dpJnupohX-IBvZXJW<1yVXe({po{
zjrU>P33nAOw|%^@oT{I=a#p>qn%OfmhGxT4h6>=Axigex*B;wSZafLI8Q2c?6j#j5
zrNn#GzR*!qvl~u=+pqj+;7x}9{(i+1LYf*Hk=1Z9Ql~FYZ;bi~4NbZ=z3chBlr267O#b+i=S8nh}FixVle;SU0LwFG>^l;q@m7Q5gR2)A_8
zgW_Uh4;Y1D2yO&j3l=Hf^(rId5&(jnTqU+TB`bEtZVn}rlV29~gidpXsYLNxpmR`A
zdCQh9KpxL<&SE$T%)V8Xl+5pz>>^1mem|6T(qp67NhE)l3#Cb9E+JY!_H}f2ib_d6
zX)VTZyLFG5Z}4mfKCm%2XOF0xnVE@NavmxB3HR||WmvuSr0rA_jr)M-ao+I>3AMMj
za>Ib~*fERjXL{KeB&*C+hQ`G)OyUzV1^Q!$jvZ6x`!1s}O8it~_B8;Nz*jW1wC^5#
z?drOPt4m+qgrPVP5=9~y8vU`ibd272X=0m(5b%}(<3^1A+EKUgvtD-J)3qk?p%E*uIu#iq0W!S
zsFNSxC|F(?o;?>GPb=x#a+349oPYaWhgEonTN5WsG#5<@YXcypG6@+vSMoxzgCBmum)uk^l5LPBO3H*miS9XzC#5_l)
z!g&2#x#t=}d+w+1z{bSH#QeOx$al!MpnTS9ZETmjFcZg}&bOxqw^%KIXA^5dfL?`iq8oo$@07%)+=%aT0%&5nQLd
zi9dQ-ud}bjAIX|o2;Oo1l_2}iKokE%?#}=52YVSPbA+hV$*Z@WC|}mtw(}PJACoN|
z9P(b6viL;C#AwnLx-B@^o27CfpQe*1x(>d0uYbQxq0Il|2Y9jUIc{ijcdye#$Mls4
z1P6tyko^j~17EjM@5MAXD3Ui1FzTS2#*xKaBPF!v@Fl>eXLk
zW1{dtjEvl?J^LwERKWZr4dO9o+vAj1j+ct-Ja)Tt3#*8TUPJD44V(9Osk1UOJysV+
z6*tyl#^m($GfqWgy|v9P7Wl?TVnxF&y`A
zmmikEcdGFmR#tnfCO`WbRn==SU5u@T#rAkx+Hx(f2p{nBwSa5bq%E`-%ii%FZ4_07bTk2i&DYN(39{3KYMPs~l
zd~flyXBT19sV#+HxQC8zeraij8wV4!lw0IN0UpXn>3?Ib$~N#`Cm(+f#sV71R*X-t{r3|s^bJAHreGnX;!5|ADBm_6vTomuKDOo(f1xnL5F8ExD=
zJh2e$%dtfXSUvdc2bV1?oDw*+l2ziyfBuYn^e7i879I%V?1Oy*cO|1`T2)nB7h(G+
z2J=Q4IUm+ntS)`Ny*MuCF}-I!TvSD;pWW+pbz>rhIFX}n8{|sY)`2MBhqSb
zoWL(PDyS92%LMj{yoxeB9{y0e#7s$x2lhabG{ar?U_o0Elv&u-Y8-B)-F+Xuo@*jp-
z9h*v|wsE-%CGgS1me_Zghb820y#8y{ULh7<
z|AcSb%6jnNK^&sXx|r^WYfzyq1|y;QZaK0StaZ|B9Oc@Q&+!SkIWafgaG&nq->%{d
zZi1G?C}`a`F_Br|2hNFx1dsE?>)Ub1mPr9VOV#M5UzgJQZZK3
zT-WdKKdHj@X!&>D2|D>^K}m~RI_KJomX^mfEik^U-D5qh>`W7Jns`j~S&^hSAaxN#
zZ1?=TNo`vd1Z8?!^DNGh6tHRZ_4lo{W0Y}Gk%+ty7LcWiclz*7KrG$aV7gL+#rgUB
z%+gP2XqlK+V0+KS#k-C3Yw3;bmfq!znDwtfD-0NhnmVDQ^9hNG@;CpD^i{h=T!|`T
z+E!NZ9Z*E}l@ojiVQ}vc<6_O0(i^xtNa&-kQ+L+Z)~4qrKGz_al)LIYhDB<*TiPn}ATdl}rJ-eHti3G#b{Z2W{bz*NZ{pLJqy+5Z
z=EO4s=egj39qD*-2}}Vve^eMRUgX1Q;$%C526DQKvHkqpPVo;pEMN`8Sh0%w)9265
zurD2&)n-LLZCLK%PWlMrLNim-j#-;%Si^osk{R;&G0YcPuJXD?d)$it94~Q`f2I$?
zd044PFV|EaC&l$L$Yt?LmCRWdZ<9nh1G{_)xd(FJlIuU_;3x|w741hDe1I95K@f@n
zQk#a$!sKUmOna=wiA?Ny$C3$R
zNNj{omO+x!8aZefDAgNx
z9b#jPbU)rzcnP(v$ajE-$m*uwKjpLq+3rwW)^f7(p{zD5zGAA{BK)+Ba0N@1>AI9|~
zsTk7w{N+n8ZWrzn=!Wt33@cMpg22ZG`1<;uO(L2JN&bMB{{7eSl*+k=P+x3^g&VA{H?7BufPk~s)?l(-S5{Wgj{wY#aK`IA(UF>(
z%6jVVrcImbBl!}EiJZSruLH8D$?Ag2Kg1}=j7o0KBx9w^u|cb^xtJ)qe#6Nq{e&3%
z8VFF_)TFk1cL;XT^ciYcY)8=GR_!+5_wQ}#I|3+NW}+e^C%Rufq~W)2db+x@ly9u4
zpwLlde`x0M)!YAYu1mhj_@=T2K(4Z~G9U-UfHy01-MANBot-cn_Ve`xpyWg=!Pl?L
zXf%Yuc*z60_3wDvcq(DTH!76wa6+iYOAVKsU%u=HSctIm*-h);@3?l5oe}CDL&GxM
zP9XCm92{s)s>BP83F|~l4gh?~gB|N{@1y?T@b~He%W?U?3;h28@c;iGynYA%kCUpP
zCole6+}$<9zj4MGo!LNkThG8ik$t}50@Hu_juQ_bUo*}saMwteWDbKFVwUzGV!Y87
zjf{2i2-IG#FJ8QWnR3Ykax$*|p`qtLXvWFz|NX!638~?bh&i46{l||zd-nVg<0K=a
z+43)MRcl3-AU)R#_h}>@Ke#ZTnI_`Do874Eh!<#6OIXlmxq>I=EUI8Whp08iOd9Jx*YAG_jO8r+KhkxIt-Y@087))Om$k+6G`j57SiCjuS
zv3dOa%Yp**SY!J8UpP>Qt*YaLg;72LsXjb1qH~6~!tY<&-^Qtu@V2e3t+iEC_pR3p
z3HQYUFoSR?53U=Ie)-yIOam0pojZ3j4)H!Hp|$#{WoU?c1KX`vh^}}n2IU^FPiNh}
zb4T3i`_stE4eeC_xQSHtSUIAz1TugZw>QLb9zVW=l9F=g&Mzf?P+TCTL3v!aGIf4j
zF}Cn?yyt2Z1{_?C2Zc3RzP5_gzQB4vGcz*{O>jiSk*f+7epM{Q8T51nE8s(AWzIR?
zYXiq%Sx)ZzuDZd7dSFgpLxVVf@JtMlKxQ6{YWwC-kuW}YOdRb>eNq=
zfw=k0SdYS>+TCIXGvl5}d;a?QvjHA{j~}PE5aO5x_qiWmPDn{zuvrNXrop&(j}Ux@
zEz_x$#aY0&gzGb}Hmyqt$z&SdDhB6(pi~J!2Cty-@kXz^m8E5JL4X|FK0Z@9CnrHq
zwF<0m5Ozz#)%x9`VPQWzauadb;ot0jH3~!rQHCQ``NBvH26!Wd(Xt7_If`?2dhNnC
z1|FJ|!hQd-G`2gAG@D$z)|;w+$Y}FMp-F7O<-O~An$>uE3E@%r@&9tBV%+oC!HxP_xg^WK5}<_y$mt$qJ~
z5LDwR9kIi4>)EQe@cg5(WE%R}8IkBa(Qma24Gr~|N_;YjG<7J{<>t+XSW!(A$8Fp0
z0hpa-&+e(BqoX5~2LqLVHH-o@UmcWk>v-*Y19uXma-^lpiA1h5KPCEQumpfDV14Id
zxQo*jO;`Fcd!Rc>O{E`CuQHH&{vU7V#rQZnx^^LM_Y8+^0AdW}EA17h?}N7-4BbRQ
z;c)e1JLYdNO1X)oOo0l}4w|@r{jf?;ZDZpbm+1k-rb~T#$QA}dGxc)cb#&;N%)o@n
zLGid@5erH0DY-KmDG>7XNu+l`L12jDQTN6xK7JBNcW9^$PX=`m5Ki)6Ht75U3VrnG
z5r&GFxi2~Ya~HpQ^@SE!VNOm?ZmtbzAYuYHFg8-xppvY_ZGwngv*0D*l!?R)FF^KO
zy!q8De8$b}3$IZzNVN5EF?#g)@xbIHrlG8W71RFV4g8NfPsMRQettr`k-Xvtu)Oql
z(kepk%o$xFk#GZgD?`JtASpoKfb@+H51*Up6ae^y83Bypk?+1nI*)H}X?1;n?T3Sd
zgGd@PYW29xZIf?%dwaXO#JIW7sUu+n;x`4kVh~!9{G*+d*yrQ4jqDe3R#crsB@&g?
zx!GB5>5CoJ)uUK>;}0JP2UplwSX&c0DkX~Lx21I@czBd-hEr7IkAq~K?@2B#T>^-P
z?ZqaF_smgipntW)ME_n#vl)MsQsUn?mJN)J{kTGwGL+B1V()yx4xeQvmsj6NvAxJ)
zIPi3?{dc=B0oeU**~alVPc%deWqdY+eNg1g*!9V8M)y46)-JlDcm90&Mf=*}-=GrU
zH9=Zg^YEeBvc(Unyp@H8&8~~Yn09T(gMQAXDw{TMwqINrjxZMd$qqtm{s*va0VPGL
z3gz_OlsCjesJuR0OFySswevUj=zeH-R+j$y{voJN)Mq>)7)l=(y!uto*too`%urne
z#M6@}zxw)~+EftM0Pah|r36hf_Bm^Riow~lHOSy}UKr}Dsj-yJAicoKL%y+qWeScv
z?DOb-YwK@is38aoE+LI0x#xO$bu$UsMjeAf=V4{x+%qV4NCMD4Ej#`h5%KQ!eK2+~7VsN9LDNxqN%z=dE?Ul(H`aj0J9bP+
zP%u-&?ZlWwBjOJ>LOPb{2t?F>*|c@D$1NOf&@rF=fX#MGYsBlpJSHwa_Wnf>Dz
z7y9n(6#Th~`&}FRKb+)~U=!^)*@e11tgFB_9oNr$o{&It(d~7cQ3f#BDZTvkk9L~Y
z_`m?|o;}sak)}5bHV21<m{~jl_*b&D`7^L=?JSFJHdgrWH~*
z9J~LB&V2f#d#J>`_nQGv<5gv=dJSUY%0u6S7Hg_2l}~FE+*yW=O0myhKmx?x|fVG
zJDQ%#Zd~PfdKpC{ZB;>7wGf43qk;bZW&}4hq?}
zfgR0I5BJW+?nTH*zZE9!*$-wR`rrU)vBAvNIoLLEeW>9snVp_)7OSgJP*hxmqN4o^
zs)O)(F^fp8VX~tY8|UT@Lxh1oN_{6NXB>IVZd2aL`V_u+TsRz%HQRhC;{fYsi|lZM
z=Z+TF%6LY0`s`CMD1)IQ;^LQ@Re=z!astA_AOmZc%KI|L+lT{x0lY)IVBJ$1Iyyn~
zkHKQoo=@#^b8};5PlCp5)8=KNISq*;E(~*sS;G|31wZ%2*fV#Vk)a`~&ch(lFb_RU
z5#+}S!=fC|HIF2l+mO_;b8tjQNB=e+IKn9Sdg(O+7)lZEj!#cbQ9@@Hfg9+Fz=iO)
zJOZ?W3^y2(M7*-Ht{4DzE7l@d-no4nk6~e!m@34~oNej8B}nIX=3?HX=wvQXE}-H-
zj#YcsbaWI!poqFAI6hQ}v9a;Fgn|xf<$TBv=s5CCs{$wAEU6?~U!o0B?JTg_UD{Mx
zX%00IW7dllUlfp_%(!RNl52M6*kbYZ9}FqaL4tJoRV8MK0AqphczXhMxSfDXl$3Vp
zh|kZjLobq|BIxTjK-C4;vqXgo*#K_0!6rlT>6GVN^-IW|l&UYu5qGbRNi$jLqVbui
zn3(;0_i{cFzVvwlN2OK?zbWDYj@N9budARc!VdlbA{}Xq6?IHL-8?te>eu=n97;pl
z`zXD)Q-A;N+__V?{ffQ)G~i0;jM3=?_QZ-4Y74Nu=ASn?HA$2IIcnp`aev
zbotVy{g`FF@ywoWo_Q-EL!6jGX&I{){)`bmH!AJ{aGMpg&N1-$zIjUE69UR(D_U4*xDSuX4TG~&)Doj%Ji&wUEn{3F~C^F%}lD9)gH%*OCHJy`~DvvJT-``xZ;m
zzCg%ft?8~burnfvER<%z69w>9>+dspIuL?u6^RK6PE30ulI?%{h2T;VhBI3>0tl3Y
ziHk**=OyHQ32#`h_&>JZE9&rkQ3Wu5u3gvori%*#dQJw=*5sX?QeaNN{3Tdl+ZsS|
z@xP0HUR0#lewQ5TSn|J5mjAS>E;NLMge(Z*jUw$>Pwc#H7y#HfM}ttw2I8To9a&1;gH-%Cu-4>_9lvQ6j
zPI~gmR!n?Z7#!a(Wc$3?`MFTAwafezt7mHvHcv`YlJ*b#s<80z55^I>cp4=S?wxBD
z0Xg58#`8!4VBzqT9a6gfNpoV>+M{nlb8=?|0q^X^)bH!~-@kvkHJLgMZbb0%WQ&wa)0s@*QJL|%a4Q4*=
z9~|7x$jD_kLn$dKxf-|~Aj*K}o4)%CQB*rBv8MmSZOw`nr1L+rzL4erd
z6FzT`MAY5**)@!@g|>;zG)(_e<=dP+5zp|HAU328Xu6b+orzILj#S8Hp}NS#PL}T;q60h3qLN+o&6d(^ncLgB)<$Dp%&5`?hH~)O!<@r?3NJ~wv
z!UJY}jXXUicyf)&2gl(}=r-G~8}IfTpBT`=N51F3Yq$I5u@Cg+=RZ1G5T?b_HtqB*N!1!Z7d}ofk&hrlhSY^7y4O%lMOrb>384TjsHV1WTe+GpnhL3
zAt9lqJ@S03grwv#9v;EKpw>HbY&c1E3>=AS>z@Wy>6OZ}(^`sbkEXszsm)qtT|{Nz
z`$YIU
zq~&3fO1xB-`jNujqiYug1oDA%D7I|5^yR#onp%?UG^zcf>p3EV%gzVf1#+3{xzXER
z5D!$~pxDOFh;lTu^I04ODDkv8dEm+rt>cVZ@HjSXq?oz{N)!Z!qFQAw>OzjclG{FA
z?fbd_nq0P>HyG|Bjh}L&8)N#I8(UjL$|nx9VggJ2%(*;}xGh2KT*E^{l}Gy@+Na(m
zmi%QNScC~*Mu*pi=fb5pQK4GG4z2R6dHc2>N)ryFz%f0Hz1rMOSxst6Azoex
zz+Rv3-&)}M_iE#{9NU%+ot>TF#K^gOiyc!C|4b~BFzv5Gs|uJ*L^>P=Z{v9``pe6`
z@x;8h?`L@Uqb{SO7t3?!(W7R-n;(kdtO5c=rhlcYK>gO=Z#esNa(FQ&HhSm%#Buuz
zB~Ae=1t1$RU(`UZ1;uL$YQ~&%p(=Ih=S9{?qEPO6E>BlutP4T6px6&88H6(E71aCt
zD!j3TJGO(@fJ;wh6wwLs1IgU5bdInBa@yppsm42hgTO&Y
zig`qRw9visW*K_ylFQvMkMZ&bYrf!MvYQ>(j!xdJGve8XQU;^E3v$zh1U6Q`?mUZa
zOKH$!;_nNIg_B#AP&KnBvX=ng;&)%^DG4)zRhRoneH3x!TwMLAZhd}X)Q!hY^iybQ
zO>sL-ev@QlOMr2Z5cB>RPJ3vvd(cf`lwcI@{dS-Jdwtr%<6JyETC*``^0TD2rlvpCPS{kd$y)9d
zkO|6KorkFz1?NX%t`!7M3*j(FyM-b+e=iUQR`BKmNK4~6Ubwlq!K&sC|C
zgqjvw_*I6K?Hf1VL8!Zh+WB(NdD}=mdsL(xlPR7jB^}etktt0DUGdJNHeN`Ea+*j(
zmApXA{;rP_=-PaUYV~E@c4$2gop~^Ixn0h7Cq4d6jGzrkoz9iVD|#0-4u$^vn=YU?>hw147Qwj4NJ3nNVi!@E
z_>ZYPC`s#}nqb+m`CT282wvji+cLktFcMe(!ZVb@1yEjfEv?Yg)f!dr
z-=8Wyx>q|iIJj?e@>5hQQ9nYv7UWA`+$0zS?%hNDijJP$hVNRM`6NBF+fGR7s|fT;
zB8PD)#g3yYVI(&Q5*0W?>>o2h(9xG7uB{$Mk;Rv;VhWb1$yy*9-7C|5Z|_Xx>&wem
zAq?d5nhVLUlVndFVwr1@?bw=_JT=l??k7v9r3P;S4$Syj&Hs8J(EDX$Yb9;9AjI@Y~!R*+>ny$Q0oURk3m=$`?aG(0mR
zjVY=0nx4LXLnPm+L=Olzh}=y-e*_CfK~QV=dS^A6S^&zjQwx8V%!&bNB_
z`P4Pr57?Dh!=E}2`}FtqO+iQe6|z2vvBjeN-s!m8|JL=R*ELcVq_N09a%*lPO)-7l%o=wzlGP48<-FWeqTR4&Qc`H!&XtsL
zkdcw`I0U~)P!#_@#C|yvO95&yQ#r}jeM=iHYh}l
zR@C-G(hh7fkiGjEa33(hjc+%rtE*F>^&HoIG{nx>wo|On2aMJXBpF*XGpop{V-GKa
z)Cf)H0+_28Y%hF@-VPH2&IN3}bBM%Cc=_9q;Vuuxs={#KwNv@{gKnLt(hoevqOR>(W+o
z41BETXz2XHLRwn3liZyFds*4HD1vlmJCC>7d@Zt1zr}s*7!xSsQL%2M%LqHjVsjxd
zW|lYvp{lTOxTBhO*s#Mw;iWH|bkG>J&uSNy@@{Gr!Ept&E$?x|L=PJ%Exf?7ZR5D`
z+rraxjgX{Vw6?a!IYO-Asmb#K;J7QxT~7)KoX4{ZJ*w#>rA~TAF))<9xlX@z6TqP8
zVd!gq88seM16<2JgjaOkuP_OCI1dP8rV
zYr1*qE`(2bZ+gI$<^=iJfc*f4=pn_CP8zU+4lmcrmZosb3xCbtxjOUnW=4WoSTIWy
zJ3u`79nX)W)OXL{p8%Rmd-_m_2aI@vY5Jr4Pw@j}qUWvVD5S>7-a;1(2|=~)zBrZ=
z>lW%8CsObnY{Px}zII}!sO!y}TQetyhS2eKL?Q=!W?HvMD=gj4&aN+%ot>RhI{Y2L
z`Q7pd$VFKzI!Yuh2Ni
z%LG)cVpX7Kup)L|tu;imP21?#
zgE%r}wUvtv8}<9L0GQrw1o!P=Z{J@zGCrQ3zKUos8iu3w=(62?=&bBGI#9I%7_sfz
zsoV<448dP}!n0c_HjR`ir)e!hksaW{;%&~q&-2b>2p5g25*4YdYgB{LVO2Bipu9BJ
z{pv(YJQ@$jDfNo5zl}qK=f%ekNZ!(FRe=ch)0kYT=Qf;#p1+3z7^r)>_O5G>pK9Pa
zK{HtHhRb4{-Mx3O%j)vY^%NoE*m;iU$Bwi6vde&aMN45R5l1Tog;@2Z+-X^^@qVwsFKhO
zaAtlNCRhW!AP?>IMS2+kT4YKky&J?okgD+9TsVK!1fGI!`3BqBVwdRz(_cFik@}!R
zp!qDXZJS*zIQWs}f`j|_mjV(6i4_ywDIrHX4?{hO4bQ;(#=ryuli0`TfVJm%nYssU
zulmnUvpur2-0SWC58R;Ye
zf=Z#H6{DlDQ*ezratIQ+2~;WQ34=uMj^>LIODQq2=zI>}=-TD8W7rh*E{AgQgAli8
z>e>%9c6Z0*Z##t`ySDmQn$_=+&qwn-{8-1AErJ39G2%B~R2;OeM-t);y>{bVc(424
z9eckdEqIfZ^(k{
zk3%hK32;|2f?geHyoTr=?Byf!i-F25ZY2jbh-!|SE56`H`tgL=xH#~)?I%whKfVgu
zbm590^?nWsipIHM9~u>vi*2duq{aSEJVm(w6?>7xC)xh3#_(V5ERW^sGb+Ofn`!AN
zKEIoELvRZF@DkNe?fq8_OPtOspYqVaw9NzH$XX7K3=c0r#=P?t$Z*B(bPZ+ABXln=
zysqv2_8VLopk(HoJHPE$kAy0)L;2P2-UM=_(`|1|v4XT66j3@e+CV+}fmbn7
z76)5W0eFI1BylLphrQtnQxxTvvm;m)9O{S)K=-_ki2kZhpxC*&8SI<&%S(zKC2`CVcd8$SMSqq_j
z$@qS=k<~hIvOUqr@Z;OJZ?RQtl&omOK94|*MzUYM<;?W-h1kXr6OXywMrKabQ^t*Q7Ut)n^e`-T7)G)gdk6p`
z)Ceg4iS4^V9@b|j5W(Y2hD@-k%LP;XksieGTLhwj5?uk&8rO=F$kH7y5m8Z8sb`=l
zL{Sd}nD{uJ+sQhO*kg;9$m5Q$ZfRlKjx*
zWcGLwYC??dU$kQ}fEJHa@qA}@w`wHsMQ9cPSV3BXKwSXKsDmL2y?-AymwT>pt)x{R
z8t|uxPKKTVOL$>2A|Vw|D-#lYFgf_4_{mt;(IWP2R@Y2?!Q*CUW-9&Q2Jx
z6GxGDeoH^_aEIUE+Fc1qFJ^cB{^DaLz{h6;jD&LyrfCb=x4`v|n0QWC?IB5pr+<2V
z4TWq)#lVX4+gw5+)=zCpQd6xfyf(zXf(*eE*mPX#56*)0dG!od(*#*v`xZeX)N^JJ
zAHgQbh@7K+R1|=#K4JVO)7H*U2Eq0maI4~W1wl&he^Qdr!A~*Y(L0C61qlokICcv`
z;49AOl<_0>vsyts^N{R>eY?&*+}}Tql5P4eAqcclyv8m@K54e|o28&_8+>E)p#;|v
z*c&r5h)h#gl)jCp;UmPgzR|k(0t1o!M8w7#0#4qvfnJr!E80#c`I#*vPg&ml({~9i
z%V?=vIM4do05m)+D{HKzA_|hOphy`R4YK@H(53WwwJPDg2sBQFRy%21DPmXphH6v8Qe8*6RkM8y!2-*
zA5A?$?=GQ~jQati+T!x%deMXmxjX;Vu9zR8Irfri`4lR|PC#GixMdV-6HU{CY}B&x
z47xBypc+BOq1$VZkeQL~xNA&U1^+knBSLUzBXnVkAQWewZ78Jw^U1G3`ML{U9CDSh?WcR2HYMmQOZ$nR8L5XUd
z?}X@I&$&i{7XcGL22c7t@S&l6c5bV%7M<()Kp58;aQf;&dA7kQ8IV;lD=#8
z3zN6sa!v+P-oDNGxQW%*=SlEr)ztmL=fPwx61{ttqP~Jq;m8YKigENv0hNznhxDep
z5y>w5n@f;DJW>!`G^gFYyZZjV#7ytaJA6qfWQsrtu54R2oR;6Rl??~qA&bhK&WXtz
zqRR5egHo$n0u}>9eAGB`ph^!bUmSYYY>g7-b+dwJM++~Ac^?W>yfM+yjLeoxni-^-
zmzURgSk}=^ibvYK*+Od47d-!&pUAW(m6d%(iL;=9ju}MP;Vbj#eo;l5Ib5rIa4FHXAg@89?C-;cJ;
z`l_lV$b%7~XQrmqI<}Iy!cNP{AtuCD4}#PzxZ2azPP_^k@(aWX6Uio@SK4H_ow}8Dg_xeGsPY6
z1);VZESLtix!{NR`dh5OV}Q;LtUJnQ;E4g-&(E4iDKYHd?~1(*ITi*8Q2Cl#T7Iz1
z2bzJnFI})8J8UnMkWkK}(*m8gJ|tFO64vEq;qIQE9w-gtdh!(5F=^L7FmS-3nA}?i
z=SEvwXoKZn1W|xa0|Ra8RE*fUIX`s~9b?cqDX_P+wRpq#x+<5a>yPYva~O3Pd5sf=`e}
z;reb${(pDn3QOYpw}v{PFoAcokit~ORyCmn_q(7NxV=HT!f>?2O4
zT_mt0sH0+|yS=KW2vT3FeWvDK}EWNGEHv5
z1H_DeDtUY^9&fl%weV*_;cTP!G5kQ+PX#0Opn-Aw3q!r=OA&^PLVsH)<
z3jKhXbgxKuxRj|W!i5F-`z{hF3M}UzIJSR&Ug3|$W?75Y<@2hhO5
z{$dJ&5w#nWYSc#SA_#3qb+1K!C-7zp&M?dsJN
zP9&KZd0t-Llw`43=a%`AsMCp9sgJksK_En0V14oh(p*)e*
z+BX#ji@fs%*50q()>b9sKk2#~&k}
z(6gcLw2Oo=jo_B;u!^-gZH5!cq7~?SY7qQcSCIF2KATIIGA>8M4e6aG&emy;ZP4#<
z9hYX0gP#hiA#=K@iHlt-ok2aC!)
zZjlrxZ2(Tj95cX3`jp+`)|~aLsW%iTsKH&nt-C;cC0Qyh@4DfOSN4=k8IY3Mr@58o
z<@B!(W__x^J1@T+kHYfo;1I@n?rg>~Cr6BW(kaGBvpdYYb@6|FczE5(Ny~WKHQ~2|
z)2PPgx_WTwRv!*fDBpC`)wQvu^c+6_{lrpE~q7Jp689`UbM6tf%;=H_#EM
z4qjRS2q>Z4K0NdHd-KLuT58HRHb-p41D=4l9EP+CEiAxk;U%rW>n`4A2gDV<
zUMe3of^!^|BCB|~(U$~-#&hCC_M{AM$L+gEADtBZxAMFcLABqZmqqs)dX!RcVj9K4
zUsszl^#$5PrnhROg;o#pMb#SwY%oJ}FH&aO068qxn#YGR!U*R^!ONH7HRho5Fvq0m
z)@-^Kmpo;&bpf8|u1|$|dG8)dyDzq(9BrZdNP!(#V*Z>dmk$XKSY7>*9#3O^y&DQM
zSZ)AExqH!-C0HU$X8rD?IX{&zlm>SecVH4e7nIy*-2Iv;Q&peIa;0^wbqBQVO?ug_
zb4)*9bvfPYIoAqdrAKD#J3sHjGZU|v1U+!m$BzmPyU!^qP7QK#F9huNKk}KEVmnlR
zfW=qPCWfgDY>#{ZYA}b_ozMR0Cf1?v-;qO;a6k+cLy^ZXC|DV7h+bG?o{7Hm2WT<<
zk3ar>L)PbgKYY(3T&{25vnRsKi_F#4HRvi0S+p>43%aJ(4@+&9_#n!Un(>yw%*Jjy
z&-ht<_@&L`4fZlzU%0YZTHo?6Pi)(E{N(v&PEO7a96~RdATwi`&ID!}CS&HJEoq
zM|ZXyi*306CxbVW9CL@+9&ri@6=rVtk$C!|`Sa(O=uQ<7$e5JDlv0k6E4^>xss~wp
zKutDr%wo8%6;=kbrU3?!(zDBU1rt9L0UPTbCeoM291}
zHyJD4zI~!7iW++?_%>MLz?*R?!@>ErUwLCuzs;bH;sz(oT*NtUZ*)7YixIPfavGU7VI8
z`A2fdNaWt-GXW@o_a2@2)z6uo0J;XRu3ouxi94jG$nYu@=ispTN`26$|M1f&hy-t!
z5@c7%)O9lTynEy?T=<4xFqCa-V$%2XXV8AJbOuJW*kl_PBPu{I2__F4{A%x;V_}L(
z8e_;wq46eC)b?ZRIT@L?Q(H=tiHzn{yHD(efZ*|Lg9O7|PTl)WR&GA&Sr(=?rj}={
zI-(6kV)8qb1jI|ygyIr5kHbmDCs>Ii1|$q1G;+e1X4N~eZWnt7H%Rjs9{-EN5h&AA
z@DE^mGp0?-bQDxnq|g*j@+Uomx~mtpkXCyCNB88|+Cu_%PyO+WuJ)qA2Qj?hFDfm?
z;Hh0JERYr8;2|_)NW&n@_i2I_8H0j=VEykQ54lBjED2wEx8M1(Nv}RxvP=XVM>LOEGHyc
z!j32$rk6M#Up>6LeajUiu7=w>`py@L>o--DD#wp{!72yOw)!=vls5
zRv3dySzDW$e2cyKIrgoSNo2P`JyU&sjKI^H%}oVoRaMsr?P)^{#!YDT=8}<-D4|^%
z?}eZVZOcGTM4IwA~Ep4Ay28eh`e
zZm~8A6l;r+id?n>i!8{f+&Ko?0eTqp_MD*vKEg;LVBl~;KgFbVQAO_?%4evzpv4Yz
zKobtW9i7SDh$9mJqkTi&AQ~?Kc>z7l_~5>{D~p?m)@0TbSkOUb8Fk~FNb
zaN{Ul;44p6vuy^lDzdUK6(Wxf*e@(^zbNlZdbEMHq1}(|qJ3BQpMqDt=6+ZYt0rbJ
z9RN?1NyqmdIBsZxM|*Hc03G~(OIsF4SnLzFn|oR)eu@2cRXM0=;|_cokTX2_GDpU#%ej#
zU_03*dT;e%j`%9m{%XDWZ9$OAA@l0s%iTg~LBJsS(D3u|rRCAw?HCE^c)1Qsyf|c5
zu7!2>Fs&8wjrP3Wnbv&3{!LK=xqe}R$&DLLpS5XX7_{$M$|mYJy-{aor#F2f5`QSQ
zjVW)NPiMmI1CRKB%C=T+@lmsvJ!`xDh|t|r_kFHv+lbcWB<9CP@w^o63EV+_!i(a8
z42ubW6mdERU*wOf%yKaL4JozBDkT?&=}G6@eS{#e3TiM%)->A(sv6TB_W&zkxTi0)
zA)c68@=i6aYOlYplB%k1mXKYsare31s%SQ580i6b(w?HqY>|Sh{oqf5((6C&K0I$=
zV7di=n}w#&mzC;mJqw@ABGCZo#c_IN8#N{C6{Ga-KH;l(x>_g!*ELY#kX}pf_l=4>nk;S-wkQ0!lG%)9Z3L0OGVTL}0r1v4_YqAPU<${-2RFE*
zSdRdAmkH-(?Dwl`j)13^tgZF5w0vkV-~zOuMtECrSlD$)1tu`bx(FC+{yc)sTOW#C
zy#u>;c}K}U2B=0&O~_dSiEx5l=d)8=kfLNpR3Z$QIR8vnN(X;Y;d>W(ocH27Vo|=&
z!7a!s_v!muyWhbF^CmVI^`-ni4gThh>39)N4LpB}3;1nyoNUVm(Ttz)1kq+S)+Dp!faIzZ5=y
zzR6XTxGxNSBO}S#*$-cX2JsM%QT}bMgluugjt_2+zB`Gf9ZN!?)z7aQTZM*XcpSOr
zYhdSylVxtm9r3Vo%&2_%ZQD{Ad)%u`;$7YG%VP|UBVvoV->IsYPq~LjPiFj;u8z)!(V1RQT@YyEW-)_b#PEZ5
zy=CwN0+#g^T;0S{#tUEm3c4E93T{AcB2==0ER`FAV(7tFS65tpn$W7{OX|m!;NV2w
zU~4-Lpb{_Xdbi;+3P4~GFG+eKwn?K51WJJN8APBUBH=VntU@Gav;jVK!yN4FU*EXz
zKY7Q_$?4DZH0z6}DJeJwMKWhBFn|{Y*M|>Pku9N19E%z1R%fgMlrB->3jd7K;?S#B
zh_gKx{~YDeCh#c-hb1&u_;A=lveViZe?_vf>?lJ?jR{N5&AWt5dwlBM^+Ed6y3%hR}W&m
z;8qpcS^rqg`RzKGy)(?Nb`tB^6#R&r6f^7dtgN3{*J+wZK;#ZPNW0D3VM&1G0MUXj
zlu@Ju&#hv#X&D$G))e8X>-zA)63+t*3RQ3r2}s%$&YjzRuL?SQ(_9f2mcG)j7tBmh
zyF@?t^VYGBa?(ar|1H4lKo_qJ2Ct#vN@pHSs()bLV*lv%K5O4hM`Zz=3yma>NXq|3
z*>`|*{r7wSN+hFf(jt4yNJg?ZnVH#SRWj00D9T9oCR;`lp-2>xjO>z%6cwo`D*JhT
z```CD&v~A6o$GmAUFW(_cX$1M-|uI<->>z~#WoHPhq(VCbkdIIrmyqB)`%c+q!|$m9O~vo|;{d^kY(D&gVK>@9SX2~{IU2~#fKTdC
z!;O^35yxAw#s+}Dm#m?wST9x7>J$pKY!19lx|=E7F53QSmYV!Y2TO@Y3$@X0i$tNd
zHXmc_&ep1`aq^FH?yj0LHAN=XN*72T3D{BKYNDPvhW4V6!eyoLqih*iMkKQk+(fok
zR&p0t{vaU@iW6y8NH(5f`63e$(XTKIAx=bfexoqJNl=O?zYC4s4Gm*&-aH1)sOQ{^
z;O~#|vGIp4ANgdAsQ_AwiU#4#z(*WUu%H4x
z8xtU*!hzitz7VL3%`JCO3ns~dB7(CIa-GhmvckdzRE$9UXYpyM4$DYOgAMcZ^)+Ep
zMQSI)B^h__%uGng_qvX2tfHc#Xxcd|E+yq2?1T`DI_S970GPAFr^V)z#}^%mL_R8o
zF5<_q0b%U}EHeO3V|PqScJ|>yHkAptI!GVYbIF&CygeL=+0GbdRYioHaDL=RY3%0;Bgws7q
ziyLZN)c4>iTUS;>C_2$`!Gpn>@Qk0S1B)9~bGN-X;KZ$PU?eA>p6cg`)AGZYZd)Fr
z`l)ER84npYJgA93;&tKX2Gf1hkROiNf1pn3xURu#1bz;91!o6>#}R9wfB3K%KRA^!A-E?92xU=$k^x5sJ_5Rl
zCjjS=M--~!03ZdOazwu`T1RZEaV=!Z<_W6@}!Ch?vOln^rI;70dId7Mir=
zvVRrXL7uu4+cW;N?$Y|bOeU=ksq?0J?roPsDivEA`+efkfXPaVSX+FS@yr?f8eH!H
z^c46CcMthld{-h!G_)63j+kvjv11V;-Uo`h!PyBg=^z(Aqq}X)*p(tetP#W)q4~4O
z>~_6bP|)jEjjRER2O=dROzFz|llvp^S3yDntP78TF)1I476kUreEN!SV`fBt*ZCRz
z%ye6?ny`*<4Bm^7hn3Y;vz2oNhWuXpA7K4@?awf{iO-_P%`o}t6HYf&T6*IYWS{WK
zn$U0ugtNQJL5M}O^8nq+_Ki|3c!83TPw9K!5*87`jk$C=k!Alu(Erd4PwYzBus62P
zP&4rO8ReW<^TZxz`{aaXFZ!6yPsb=d2YieR27UUwl243xIaj8kUnxqcyS>4}VB;8+
zatOt}e*HP_KeHZuI^Y{NckDy-c%k%cgVLx;kag@qfS3xVS=(c7Na?YqrKL5HlCXH-
zbcYfDeq3jB
zKff77)8Ssjg{zwG?&T%S5(P+q;=_lZ*t(FYJ`7xpfBFIA7XS;mz}BS}NB#-7E)+_7
zT3YYC51<1CN{o6Gef`h5IinYg+nlg`hc}1OF;H$wrKtqaf@afRf3vEJ3a>#Xe~=^Y
zB`nPmh-eYW`cnVpBM_eu1c&w(XeVeb3Q$l`5V{^ezlooV;Pnqxy`|`VbQT9pET<|o
zwoF78U6cTHtTz;1jgH#BqJr20H@+;yHK!^sQFY+91usiq`Us5}_ApZX@GpJz$jnGe
zg0b)&P=InAJuk?}g0GM$X%`~sTYAGC{zOj}RI6m#y_&srHR-5x6qU~tL59N~_OfZ=
z+t_>OEuVK)xUkmxv1F8#Z$7m95I=OnAUP;=QEmXPGM5d)B@PLkx$hjl3~FPJP-HGT
zfEhpc1B)?wa))_ZY4lI9F^KTGxcCF4ojBVyVCnd#gbEV773>e+LT9V+T%C=mB4#7t
zCkVffng`-!opBzhwxpr)fc6qx2WWiMv-rVl9lL*uEP*7BdlHsQt?4g>b^v}&y9cMc
zJ3GPaSj)+=dj>*HXF(l|{HU|JAuZ$f$!k~>{`o*7A%>~EjI`GFCQoEwp`64qh=geN
zvhw1LVv^DYXnhWWCl8xsfT9H8(M5dXQkw!CoLF**v2FrX3%o&T?*1i7KrgS69?DERi=In$2TRPIyAjQ?f5}
zFttLv)5=t&-mGwv`^f3Xju`{xpL=KDYreyC9)65V-1EX5!TM|l*-_#nk)$InAz>VJ
z3hcgPYdD@f)Ct%d5^UFi9)j|60x1bnZTCLMR;UBLGlGF&fyD?>3a0xpH%DY7eK`niS^-2(9P*MY4BPyx@t&V)nQVFx@0`4zJdAs1!q)%^0Ki|74u_@;K?
zZWYJ6!{rWU1LY>tu7Xzn%;CsIE{8>9U|rRD0?|UpgufhgoCjbKVIgZ>-6)mZYk+3J
z@L;i^MAc6!B95?*oHqlElO`Vy|Is?kcRj7EzU1TjSxPv66L_lJw@8Xa1
zsk9@l4*Mxx?ueYM>{}QBa71ADyL9VNsJi6ce5bdf;+@(iJ4}Yg2Dz3>G0
z)1AxT`y6(EnMQ`me)Kn91=y4@61Y@aKpJ}y+Xv#v+0*C13$sveg5n5T+7$u=c#N(P
z2V(%-a6}XjqzJHYfO3jR8OB-zs7g5a0$W>JoB)OpF|4S0aaYfF@Bg@nHv(kY-D7De
z%^wd2qPZ2EYKV6=rKC;exU`z0l1ukR7=wX)!PEk@!si(eh&=VNYkR06R(Ipmr)Agh
zG3hVxXi;vu1jLJLQeqQnkPAS|E?FOLZI>Jz)~;Q{NT*r2HG*mZ8=QP&NJG#kG?5@x
zfTzH>11S^4j#?kkut93awbu9aGgwl%69WSqva`of3Tf%$D97@ksy>qwu|Y-ENxjD~
zWQuHbHm_=CuV!lxU2@8xZA+AoK#?N%f77qCk1_PxKN9I?`o%?5yCB4y8S$Qx0%|Ju
z{KMQGGK$Mam)Vl~Q91OF?%=Gpe|!m8Cs=wcO-B`+%guoP=vDCO+sr6v_Getc%mJ)c
z>K^>L_
z#DF!0NF-1(o4L8kIZMO<0Dec$7+th{sqDfCI3U8
zNvcx{pcBzX0Hmzh!%rIA!yzRl&}e1w%e!^6E_LS**}&Z?F_s2YzG@1hcE&~bo8_hT
zL=|RS&v9=qFD@QDGfUz`YRQ2=ROC|pKwof~M3>kfz~+JcA`|1Kx?dg>XCu}e&oa_`
zzz`#h>h8xcWF*6wO`A6-Xh3Cq+sfq;MvP(DkTVX;rocs&o!h9XFE_<2F_MK|DT=uf
zciUem*4mzc
z#gE6%ZVZ2V=%f}&$q@$3GihvB(75AlDbhvEN)ro_It6F=#8}cj2`U}(p_wpyj?3jC
zfS?yIjudgk5>YYDr)&(-D1wG8c=<$#MrC7SvG4
zit;PBqa=AW;~PTHq0J8b-$|sDhLV#dYN*^=<`zATfvam#)grbT6@W@TpJTIgbB~+k
zOImQBVxeoBazfd6-{$s8iYUpgA-I=h`ad+$RC1wCrupMx{N}^c4rLrl6f@D+Mfy9^^j;tkApUH;
zL`NrcM*O?1ua$aJ&CM#(WGqon{N#28m8=b1!3~d;Y-?@?2eIum8YqOZ?1;-37YRTX
zm2Q8gys|QRFr)Z`up7IhqGX~Xzne9HjObyfsuKu9nE(Cz=}3x|g@v$f!x>bu1&x^3
zaS&}R-R|9MxOQM-#CwIR7GbmC%OM|B8dPj=^ZP=VF%1DJVHv2+WfE=X{&*!RMU)iO>Q){KJ#)j8e;cs0TV
z%_gqI#a!cn73G5ugo+Qsu*5{BIIUL4N;DowPJqlRw};x}hLTeMJL5veoXUCfO0
zxw@u?hv69xJM&u>%7Hg3XciCEcX<2ywp^pZp+z90_{7W!h`_jf(EodH?&b6^H;ak{
zSLCze<1g*9zes!>%3Sny32R+1Sy0)d4=Y0q*WJ6;0K3ougiEud_LgpH#cIU*g?Mlg
z&=!hldw)jYK`kfBzrkhl9wjLe>^#i+rX8V(zj$jCw|l$>yTGCUf*
z8n`w4Z&$nGr+l`gUv19A!oYD_^?gf)#wtbI63wzsr#ATZ$*yX@%3hfxHK)i2i2y_-
zCE|5JDu|M~;AF@TLVdgzb6Wyy1^!Rs(_$N-ASt;R8Xgq33+-(B7f3o6fahJ@8B`qZ
z9|j!Y2})}e&rJzrD?qowhoE1^Z%rQO&|32yasr-%vm5Xw!U;_UnU|j*M5(@DUaT(J
zj`|)PJOa1tA3T71aP-+T?NxD7aAHNM^b24=1aw*BktDapL?zc7xEa=**)%ymEzXC6
z$0$4n%j!!xQLL~pSR4H@;t0j1u+OD~P-iZhB!2aV>J_%K(jqDwTiZrJqqu$G_EcK6
z<0^0im!22b5lnXuY{y(7TE>qzxq5(%MTpH!U~oqG6QPc!P`VI~P@FXQm^tN?A
zE`Z*qSsTC-G96)T!NV0rD43Jakfc<0DuyILQPv
z(h?J;0>0TqvmhmD_}V^N4gP$9$y&*UzJGv3|iFPMH(o0>U@ps
zJGDowk32(*0E>R%xgB+I4mbR!)c|Ji`7%A1xBcW>dKCg|gnVDcgjq0Y`r9`Hc8p-e
zHM|P0aj=5EgxHnbDe2J9Rk8=I6`)NI{~HJ+YmcrCNlZJsD0
zDBj6Cwu(zRvR!nN6)30sWWK-Qor4Q2h*O(NDwBG43xkUn8Xc4d0Gu@?A06(u^eU>w
za5kYrAQq64h}E74U9pH(8V#_XUJ{`CITECFV;dX2={2Q7
zVw$U9HeX?Wev4zJpT9q}Dz|1T+6YIzL6(P$OX2KWj3&|Z<*ve!Fx_U`iFL(r-uUzM
zbgk#h6aSDErl)oA5VWOrkZ|bKb`5^Qb2Tj%cuvp~CmmGOmJceZ>y6FGh)OwBY>UAj
z5s8#9P*`CN-N3<=FXV{0yh3z%r9s4MKwKKj$xs_w+@CVu@09$Pc5RHlj1iu#YoUy7q-gqKw+;T0EN1R4QGwTUj01{G$JM2U`J$+GjGYdY?6YgC
z^)RaWd?YD3!QvHp)NzA*d8#ecnNXB{{Pal-BmM-F02Z9aNr}SkYi};u<;U$9i=(F(
zh$AFfUk(Kr{(E4P71LthzJJfg^@m4@d+d3-v?oqLApQQQPeX@?Q=a&rxblb7!I`{_
zkh(#;SHKU!69CFbtcIVrcR2Wj3rkY9PRKRPKohrL-`t!CwZ%Q!+0*kK!#q%^0Hnv5
zT8te#{bNG=*7N4Eu_pFypT(Dce@9WXhA9h($w9d^{N#x`KNHF!f-l1jZShPMOp>8t
zFBSm~uM)3a!qWNo@1MaT2S6L0Ve3j4wmm2?I88yH6ga%yWUQ)en1ECmuNt6BS>Rw_
zNg?eV%YUeYIi!dPUN!oeXIz4p%BdK6ucihGy*qGQJa{0_+r`Ynf^*b|7&{UdS1^P=
z1BXW8EnZLblLdYhxXrE%Fo(^b8SCyA!9<$0j-32_Cs)@OfK1-!sO?<@3Cvtd<0tp_
z<;$07I6)ADlzsYvhnE*y>?;V1AHQhPVVnp5W%he1PUiE_JZsUCF}c9KFr7l8SuT)(
z%m;K{CGIl>pW=oB&x|`7miAlC9gVn#fe~7UiOZp5J(~no5DlAn4)z}yIQ;m@9fSe3
za}s|Oq3I~d;!|-0qKeK{yQ{mq>_$O6(Mty1KZUUaIqsA6Bq?uHY14Ze#>1TW&IjK4
ze+Om;_SBZmp~Ar)flMzoKHk{E;tXbGfr6vsgY?r2KnvLI>GM+WkgS@Arw=+0Uj-a9
z)=t3|NE=}X?=4Pc;Y8A!oI`Uc&ZF3KHUZOv0tyP5_iJ`bI73iQ%uI-iiULukNfWG+
zyMkb4EQ7xJB5w?!k_U^0DjY8?zB+pD=T8Dw#KORz2a`nv1Y_WbR~BBN!)8EJs+7m^
zneb-+JOF7A^kyq9tvV$c%!{yP-OK1L6A`d@%$hi1w>BsDtpd05QY{24_pEEF<$0N|eq{
zPxScEaHvuO46(>ZM>G`v#KU|ihR6X1FC6p9={4sU7RrR_igmU2sGm3?da=t{
z0!vjMBUhNWk073B^NWwk;m8`t6(W9;J*?XER8%YhS2t`zD@w`J7c88ejqDj44xy8lN6|DEa4=yZ_ldeHojC%d)
zq!U=EonWC-65}M<%<`w!X#T4@=9~t5;?tL`Naw}UZ19^hzOdQ2^WE1u_-(eiFx~+K
z0|sLkCywct#mo1^FgTr?yMVo9pOVv4jHsLt2&Y3Oc5d!{7-UY(CgC;@P7~X;l^)Ap
zG)3r;FYqVgF`X)aa#!$2#Pn`By3Ph22t{bcQ#>?N>Iwp^e|49ygwhc`3{G2PzCFtz
zN>Qr_2?%g*MT_>qR+bwDjGn}8YfPpTSkKc*)#I&up4^CqxAC6u`9hu1lVV^wC(k11t@Wn6idVD#ci{1wJ;|0+BRK0TGLa8nO*yw
zD70PJh`vdXOeMVAb$MwCltt=+Gx*dlDyrSI?8LL1eo^8T1|>q6U<>+RC>4rad0#@1
z;fq;F@zKqA-~@k^E|ATfS5GS$hb<4o2D1RC65~fON;q~!aUS}(f>*#Z*d|EU%e^6Y
zgWDxHcY&HrCg+H=Hc$i=_xmsm6^}nLzn0c5yv2o8_?T}|zAQ-28Wor@lfT|7q2(q?
z(M^0M(W{^jGB?_%+>G)@Q}gP5db7_=hM2JSXU?hL9w`6d^?4y^@0VuBRizEoD6trH
z)5LFIN*VHQsIf;2(HXwm7sGydx#4b&BLg<)ajeV5FoG5CaI&2Z}TXa0nT!A1Iz
zu_OtqVg+N)w;G^95xqTh1>VCmPfN4sFayz6za91^6-WWX1
z(R8&{!EwlrlZPj4{=T)t>BDMjo2Je;VdF3PP@n}rtfSLEFT9gEW_9uq6^<_mvo&BB{E8SUVcgZlYn4N+?oWH#T}dfs7++o3;FT1o_oQ7SOR_N
z^k<-=t40nHNw3&29ivI>{DU)h&f|?5XE&ge##=u+Fbhpvs-EkSBfzVHgPWRifBzvA
zr==CMTGqG4!nV|qx1^n#7B|foOfD-jPh6kHSzGWJCm8V1vMEHVc%TZ}AEps#U;)M_fjOAs321J#_acVd^#a_Bh>V<_5YYbXtIXV4~LX4WnPeo&X
zF&B7ah(|Nh(~DhN{CNEx5Tm0w@NgACb(H)3ATAeZ
z(-9QCZCeb)-7n6}e>Otl3XulP$TY}t{!W)GL=g2idc#V#Z
zPIpcC)ZrH0d%bu#x-*CgDEhR`By>y#ao9*DzIiHI$Hfv0NwIlDFJGQ~c{@vkHC|?e
z*{|{Tm-?p56yzH>xVOTdaB&2Nfypq6&{(y-TmF6+XqvE4euHM8pP#|@NsnN7v@HrL
zT=;>Y3Dr2nLXdsp{~{DdC$r_l+?B1n*!S$&mUcnOWMe1=saU_d78~vY?w*_JKYAel|Hr?D-h>=x|Dl7h7dq)
zH!aVSW5D^<9APQx=g(v!c&h)K(W{F|6OO@jrK74-s9TP9r&)Zl+&e9=Y|NjCYpl$d
z*;t4cf-R5No!W)QxrSO+`r~@W)u4t4rPK{@K)lFF;g&)K^fk79D*2w{?eY@*4UZ8p
zw~B)hScpmObv<4fo$*7Eqi`u?WW0u62m-KApd8G^#NS?VJx1!3)=u54
z>JG&0G!3DbnZUm2?fR$un3>lKVw$_V^|Ly*R$AzJ^bUgo_E--}%*xW&X*5Zl$Hddr
z*jS~VC3TDiu=+}rTuNN-1&JKppaR#CU#t($B!nwyM<7y<*5WpP2qo8fG;A-X7I#k_j{LUy)LhlGq-
zb3LRo9vR)utz6X}3|r4(XLEh!~E
z_#&)zbIuEt{wNBtvAy0gpoVN2hrj%A_p?C7AT(6~j9MQfQ*-OppNM=0UHj;fl8H&Q
z{7Vm}7L?e4s1Hwz;c0{A|B3gCu&OR3_TxTYcyQ_7y?g1Wg9SOZ&_b|?{?_Ecz?(M;
zz&Z~6I2v{ZOEIp4&Xi3GNOyeK_}~iv#QD>hB(J45`pJ0w(5db$4p^#&5(OtDk9Zi*
zX->0YJT4F=P3E6Koe4qJMsLkT>lrZT~;A*3;9Ct8+0H!^)ez{1ypKh9rbtEe)gD>mGLZ@Z#E7+Lpevc
zbd8&mIfsfYU9{zSBG_I>Nth3)kNNIe0*HTp7!3SIx
zrp^NghFe=FSB0s+Bo#gbWDF$6#pQM}<)1h*j92e&I1r6IgICx(gEq>xZqQ;6
zeD+~Kg;c2i@Ffabt*_+|0)=QXe9Pa!l+u0A36R+02SckDLGVL-$9`1Hc8^ihWT;%$
z0Vb$oH9B$#OixaB8hjcq0Q>+Ah1n9qupls1D0#Ja>;-ip;oe^WM*zR>CQ~jkDUhSN
zMB0PZW*k#)fhg^(GFR5GuoCta_B2#>+p_d;!5uE}1!uCVn;yU1op?&_{79@5*o
zIrvH&m>FO?dC&FI^lAaS+PEM&bO(eowVQPf3>wi|o;xR}0&3tSf~#?5n>SriS!Kgi
zBJDThPU=)(!6edXbW0wYh~R#O)Kd=qt({OT0~P2yV9-N#nwfAE8XG+D$W>M}h!(x6
z2*o~-LM~cpa@Ai2_Mc8SxLH=0b!6bJ;Vz5?fZs-CC-EFq#x@yO51l!2czVhn7e83DX^cD}val*>PEq7bC*x)_ia;v{p3zN4!GYdC}c+e&O7VgOgO;xw%+*
z>56M+V_O(F2jR-^*Ft%WX91Cq2CLasXx-v>2fTj$8fGcM46!DEbCb}$i>!tvrtt3R
z?_KQxrlCDa17tWMI~n&w$EH40Hw35>?5SGx)?6B7V}Z4*pe`r+hXXsWrzdE~O0WZnKXS0lC{6D)GFIM)gM$W&
zK5^{Do1Ae}R30C5AxFI;b$3K2H7n}^M4j)o&|d-qE3IVO0xnEDffuBI4qz8_`mOi0
zD6*9j9{2UdaZ5oKV|Lgr_OOoe@H9DzT>us5qis*kJGD3dON#yPo`feu?Y$d34IOu`
zu&%UTefB4ZkY$e!b|qgov%QmxrbV!k
zv5`+juibG}gf&7^&?R8H=*b@+*tbg+uN&bBsqKx^<1K4DvVxPG`@vOwV}UmZo@til
zajS7!d}1Q1wMex*+@Bs&F*f&3WU2k=&b~_&a|7z8_6`n}AZP#_F>EeN#*Rbh66sG0S|{19Lunf`+q92rE4-x=d$iu6g--L99-0MrzMm3-|VQ=
zfb=cb4;WIu8|4H}EDcJ~0K_2s#2JxomC`v>J>E!_YB+aDP*A6)H^-f(^}X$4zunIF
z>e>sB{A)KpOQClcm4Z?bxpfQXcWP>oRZBjV2;UXj6wKUk_j7ldU0lIdN?<>3y4_m8
z_8T!#1G@?JUd_k!A|@B_Xh*D!TnA
zF)5OnCbu-Z<5r&uXVQ}~?Q3-`*-42)A}PJ)dRD8g-}XO$+dg<_I-zL?m>3AVG$=uj
z5yQg{S4<0MAeUh#7=orV+%7m-aMgeMNw0!44SGsvEG@l!=ut6r&G3MTjEwoXfPdY_
zl7u{%AAs65r*)hGr8V`^AW3$Gkyj72J+pi7*w{J>&W?1)(!=M@E#VA@&Vp~9B}vCW
zAi(_eUiUA=7_T)heU>P7pDlz)0*h`)VV4sk^0MUNe5bFM&|=LYFbvRBS4U!%S2l%K`mqL9G~Wo&wi>>FLVtuWLL6!%5L$C~Xs{
zUIDs;v>uqTv=tw|fvqC+(?KGklq6cRKj~XhxxGtH7%ScuPqpuGW)HpT^j0(JhT-6ce?N;bXWSp{
zH)Xl1aX~ud=?CO6>{@1k5{2*N-eNihvf7S|8;C5EQRB6ei?mK0--b|0B}t#+P^{xA
zw>c&h%TD{E;e<+-`o@hLiKI-HsEErraeO~^$)27`E!sdA5fifq<9n)}bD(nT2&Gvp
zf1t4kAn(B`-vhoMuW6{LVgNI~bOaX^D}+AbfULAM^FX`zpjTZaJrh%6db$p~>FFkg
z-G>{`@~QB1bE|RXI^dEjxY~xSS}^;Z*HMT=bd+Eq%+KG;Z}|a?Kx+yURUwe#>Flji
zdo6^MVR$w*Wm3c~?2DW~G#c*J&tXWxz>aV(e6Dnz+oaxD08ymi^t{5OC5i`5FD?0%
z{eRG59bJYyp9nJl$@HHPl2iq
zu2O*#;jzpavT=hA+U%k-YB=0XcrudiSZ*JX5{L?fd)>evrP3Hc4csWEXP%VZe}E3x
z_&05I12YKdAUNXQYk{z{u>s`%bx?K{L8YMT3LaxS2z8poR`3M@^G2AKT=r}5OH9+O
z9f~{8a&Gy6+6#!x7hID`=4zCL{s;$k0U~XeV-lZ(7#GS;L@Eu}f-WCqwkO}*SKbM+
z$j#|y{v0DfDXL6JQNxY3L&7kGl#-1VWY~}C>4ynm6NldPy1cnQa5X7h8?MtTsQy~H-Xw#-7U;;!RxLKdFrk--Qj)naqTxOH5x
zvy8u|f-X68qVcw9PlchwuiV@sBb{g3Ci0BB$9z3mseLsicU?%#^pfA^G%9}Z{Myr~
z>3DWBY8RQF&ToKq@e&I7BfDKe5@jkrrTaR^@?cALJe9{`g~gwivM}=sk@l00>l{al
z!f;dJk_z8V|9a!8BbZi+rngG$O%A`HnAl;z@S@mj)Ei*>LTsZn7*fXpMRg;{>JXQB
z1H*Cw%>m8iK7J25csN7xKg9+;rZ^6#4x#vl#nNC~#|^hw+wPEO&!3+Fy`JZ57KY%r
zv2$O_!6H>^Gb&~W!P{Cg#EogwXj!RiReiKIFc_kHs}*-$#7LqxJLn`&gDm
zGrD(++XoNd<1Tmm`zcSt!vLI&+1lTXgAUQn?%<%iN)yojwkpzd!Z!pB1cEmPE~C$c
zoDH8aMM@F&C@dtM#YyI}WeDC&bRV2h&da~y5$z#XfbOo+m{X_e8@5aWQ17!u>476ru
zW{@n?DJIiw*^>K`xu>%;p^-J~h8PHEi+d-I>gZ6dUIH5c+37zn+<{~9YI(|O`@*sK
zo#0_W)s<+PXQry^1W?!bg}>0)CBIh1Ac!>4P1af2D}GD9z!{AE7XW8z))f4aaIE2U
zfd-1^YRh)qeHj@UC_S*Mii%vZH8FG&$A>FMmyvEvPZ3u%A+SUL43TmW@|I&i|k^w8Q>)F+=bzberq3aJY7*ZTq(
zd;BRY0BaD@)WD8`jX-yRVT_Uz64!Sh6s*0k3hf5u27>Onnh>}m;5XF?;{ksC{*9Wy
z3dC)Bc{w^@d>q6OL*5NDKJ;Cw`e5Fp^p7d=-G!?Grx)WPFZ?BlAzOu-OB2~L2QBOBp5J_PUN$w}iEa^t
zG{d`FHYsLfxZxNu6e10|eDIQ=n%TywsFf$iAGNe1^Dx_P1oPrBzgByF0pl(1L+uBY
z!o|xF3kZ!j*mtM9uUZ2>5qt5KtBJ_~%7M@HE0V-pk>1T2fbEjl(EDA#FMwkEv#@^u;+P3q6l5he&Ji-Lt>M|9U#}BO(o_=I^
zyAq$1mscR}3RJh^hpS=ijR7ro?OI}`00uP^hEih|#tnXk7y_;IltXT|`SH*1Gpf3h
z+16bmofS;UnYYfFnI)SWSqpqT6%xB~v#n{$AbzdnPZShgiCeJSY)e@NUFnVtKN(l|
zcNL*Q@1%eD#Q*I7kbSUjDRDV+#Y9ibF*!VsKT@JfB6g{Q%gf3bmt(W@iczSpB7z#+
zZr)~>icPW@%nFYs0*RKOx~#gEfDwvbsJhfPXkEg%To4^t7(&0??_Prv6Va|NVm~+R
zI{FXQeh5;Ny<)s6h!Fpdk+TVtB&Xo5o3W)nKN_}TFZ#t>^lSah-VwvM7M88${O^6D
z$^@hCGl!WaJu&O16_lUN3*-Cvi2C?Z5r;JGpSz1(59*2R?Cv_BI-SvekE-6cShQR_
zrbjQPr>h~GYsx9*h8NDa;7udMajzHbgO2b)gODS;J(#8;qGhQIaG&j@nX
z$nGBKHL^1^8?X@oXhEbwZZ*5G0Q&6UDi_vUw<|H*$CSt5woa_iE?MQ@h^j{N?U(!*
z#dyyf3IKZW@6?s5Al40Nt@iCCVe}9N&wGErXyr>Wgx|I^8zV2+V(`+55dUT<7d7<{B7lA
zPnsqXugdEHh1?5@z}6^rM>utNL|!emcUJ*q5(w_a`e^fOGP9dVb@
zwli7AQhiEBi7+nV!J+&FR*K6b+?5BDKB;MF5C#4MOkJz3IIfDm4Ya72;|90R8GZip
z1dEY1y!r|uFa&L0+LxgzjLw$$%dXIU
zJez=%NbruC0
z5WgSEH#dB?urRKbxfit(^zAy3@`E*#{w|A2&G89AyUxU!KlV&M^DfNdhe4S7W?Kt6
zYl}LI)K}ZB=cNW-e2Lv`J8S(%%!I1{gJiZ%4}UjL3Z(hGO6RWQ0^I~BBrYKpZZMPCzx6Ec@mXDn2>AI
zo)(bT#3TqKsDBsYZ&->t>Xyzg_eXVll%z>Xfm2+`KqQDqvvNNLoT8uLY
zunBr1eI1>1t4sbUb}%Ls;MCMQilA+v#nC~blALbHKnBtutRraPboKQ`aLtN|$w)|4
zf#*TJ0k|L9eQ)nOHUlRs(=#&Qy+=TKMrta=5055&;C=#AgC7iuM))CaxU9^~@CTu<
zr06T%uYdgL3ZQkV2%!zJP$Lagl)&Evr6`z9D9s5jOh^dHKge6d56(%z0H~RYx&rv<
zD2X{6_`-0TfWrdoS6W(%zYYi!JR`WNaqiOWM#{;%$7lnqKDCd+VJ8og#%rJMXF5o{
z;79VsX^Tyql#+sO&b!VAr3Hj)sD{S-YZ2|~VjP*nTaQV3iHXjb3^ht-
zk{kFn9G@^xCT3MW;atyO?Ac!FQ+vm!5?v2d5oh&*@`dcrc{eNa&9j5D%=()d4HJW(
zii2skv0BJ+Sw^_h02HboY&0WeR}r`
zOhY4gSHHKme140FcJ8r)FMAZ{?T*)ED)Rt&#wq4DR6lI;JM_x7J)$~Ni0#%7TlR^H
zT#GBhdr9!)Ic24Wco6USiV+q72HbDgPzf^%~gP`4I2+Y^-Z($
zeUSM7RVOEZ5w&D7M6*R#D3n^J0(1+F1hn~qb>~ucMLmIP2tS^MzjE*TfN(o~nutw0
zzipF}2nQa~tc=mKXF<+;&a9;Sw1u!%l<{;IGV{w?+6~1ni;MjpANvw1JBNIOKKo>9_Z_*ov>Ifx_vgTxkdgl{KlI|E0R5nTWr-Tvg8JjJY9C$O3I0LcZDgZ|}_e`^6AcP6gPL(g?mLFAbLU#nibF3e<6
z{OXOXmuVe*K8#qn9Hetr{+}`nOG+sp?7<
zX7lRQQ!lhPtOxv3Hpw{ddU9`hqtTZZTV
z977ARF!cT;O5O0r7spMUm9
zV+iZyB69<81P|Y2x8iKNm90xkW5?Rlq5z}+1U#Y=$0SN^J55bdCSFNz`=A??eBlf`
zyRfh3-5aC7uy(^wHaWqMJpqM0e`Hm4^$JoET^h=OaM(WifYKizAGo_#7nuHlG~}hA
z+k?ES*=MOE<}rqnVjRou+XjPwFq0Dk?LBmKz0gO5C_!6EgitXus@lmp*xPdm>K2W#
zT*OdHD98s`rGdnPsl&)def@5zU5>xGANOjb-zm_8zzXvP75Zmx-v(+EL9oIs-UE$e-ltPDTeC`Z-X-K(m
z&4zEPB-bZIsuiNS%`DR}{Xmfg(}$7KH{2HpuQFl+4n{aJ`+ggo+)#fX007+~SN>jP
zjVlvF?|QeUZnmwK`dL>|Do94T!@8h5A
z0W|#t>cFZZ3?X}v&_N=M`(qU6IO67V_;pN7enRz@Cr#hQk|V_ROW%;VPkJ4Q&zZx5
z%Dy>0eBNGHvZ*$&NVuX(p~=|7$N2uDiZK_}&NG~zT*ArfJx%*HE#{?CDp#B}Iq4{Y
zYH=wnKSFPb2k!HB^vz1Z_N};49Tq|$oCom&3`jBiaML4nL4I;%2m5j+*!Uy+`}!BWvm$pIMazPIvyCzbH0A0irS@7{H)B^}*vw51;!
z{Jy@*tTqe^&>`m1DqX0oS(BU1yV*1LWS!lEU5I$fW=ZOA@Q7WE%+!sN;(7RZPF#QI
zwlG`mKVSKM>O+oad)&JGXEuzG&Q49$GJ-6k4`H+xJ|B%fPqN{T|Ap8W_yWc|ab7JY
zLCJtC9N`>}qs0?!L_o0fGkQZNGb5u@XgM%>3r983tG5pVkTW=of^^JezgYIN8qfM?-aq
zt>do{j?>1kyG?_J7Gk7pCU&&DH#7?nc)ncc^0d1UxkP41;N(RX%s
zL$iWq6K1EMea&Xj@#&a2J}@|Xf8f}5{z!y-XU{%f#&X6##N-xo!>zca1kf_%hWK$5
zW4U3i{ib+@?TmsgV9QxgKoO8`zCqm3OcIPkfa2kM&V%EVKPV^&GYyiemBeLafIa;{
zM+-dG%fT6r6T(>f?!o2T!NW4j4(eniCCw|Cv8!~m(6jy2d~Lvd?Y&cXS5AKn;yDo;
zYj1zY{L3}nhm^X4=r&;(f|gjj=?@weVsPojuMkeaF%g;fdE2L{vzsbI%NY10F+B!U
zRVjN5o~5qty4-syIKhms2mi$-peMtSyU@I1l>61yQcyf5Rdc+6BP4WL6aG7H9y=1h
zMxL^&Dw(?gDfl`yZn)s*J-JO;mv0`wVAe9+qakG~eT2o_Qr3EYt9PMYLRM_X?*hXf
zt$XM9d`S@Marf|~O_#`dNb#l;UTZ8-IBO#FLLbl+&>aP-oeOrZ6q(4Y7ZOY0gPM<;
zVr61x-ZHSM;~qycKOX}dTey?3GF6@u5$gtvz=$ViDCBSxqr~b%!}yvD?A|F1_np0y
zKl^2p^^n)X27gf$V$EtTIJFAx(ja1~R}Ul0G53bdEAe6_&h^ipdW)K+p&Hjt>BNu=yHfqfyg+%#)NBrQBjt+dCmKJ&x7ZSy&=%v)5Cf}1T82_
zDukkfE6oT#K^BJ2Wzr+ZN*D{XZ`>FiCEE}zgX9UL8Si4zzhk~X_dBPNq9S*(v$4Tx
zCoYD*AG2Q$s2d^G4hJb(M&{Sg;mB9B8=!tGaEM}QDI6CEEgQ%2*PXb~Q)L9lHY#(_
zL5{GcDcR4Mf)oy>9;g`N_X;MAYR_-B@Q4gO14G`(28^e|I;Cv7T~_uQ;iC`l=WE&8j>3%mwzrofQ1;Y&HKP!XNox+c
z_oEs$a3GU%bRS0&(><7HfB#+w8UjR^KT
zIgDR3?$p-q-?z_b@i#U8Sr1~L;9p19`#*^smY`16jld}22UfyB#iqJCOTRRjIl*d-
zva1*1^g=d96JXf!t!N29;($Z^r}E4bEN5uOZoa&0s!hEYv#j#uwpu*L-`zW8{{Z^3@E!43ZCZX9US6
zt?7`t6IX1=!a&RoI!a4TE{R8Hs>MNvNebd%(cyf#b?X`kri_dS_4RNS_WRP6m*SX!
zG-0aRF%YVo>ma(Qe|`in{~bC`{6EDse?KhoAOFiAcB}nO{2nPM
z|NYzR+jDaK@2VT#JYWA
zMMerfa_OJ1!$DX!@CWMZ>j@t^iaM`kW#Qd1_;3#)5C)(S__JFE)hyNg8ph}K+Wo>z
z6iX}{>Ck_^Z9|$8#{q~Ep!!B-fN2Ex2D1h}s+rkk*qVW29>D$v!g1}Mvx^JRf&nfqDL-BaW%0*a
z51rDS5|x!b3%3%uIx^4w3*biaM@uX&c$vwJSg_kra^le9zZ1lXg0*bF39-$UFareR8d->T&BuNQu1V~9QO
zUxI8~zWwmwDrylv{dD9i@sxytUBlsrX%t6iG`1*aV~EQDWS|vh=|rA&
zbb#_zB4ZZ~vj6>W5Nn`HPfbbT`yNe1xBz6sxr{K;p;~#8OOD3htoiW1;o$=ysw*w8
zK--L~ms?)`Jd8QURoI>|_Xc5VEb9KKYuA{Tl@MA$IAu{`o!q%C&dnPW=cS|qKOD_j
zD=jNSfEntJr_%q
z8aU1~zM`ZAqepCTiNQPuzK-tu$B7B3pTFPjQB3Gp@%8-)L~V+GCgZad7JO<}*4xT@
z+ze@m*Sc`w;Lmb2pIFtS*iCJ17*SG)giUm%1Ox}M>#0fHJVd@6oOr0drKjN9ylA{%
z``ILNBrw7(F9qBc-c+29)s^eYfA_5tsyqp7tB+&RC>*gS42vd_pN!VcwG*~^8&UI)
zXn=FABIrEKT)C%tg@ig^y{dbt*|!y+L)86-1Of>cW^SW3#J|bF@R;gt{}g$TLV4$x1R
z64`q*j0#|l0;hjf$kcd(O$P>
z#{xdI(E=khTlmqZxhE|>o#8QJm%z&F)`w|Hfw7Hg887xw=oM{Tnl8hB2)O@^$TzXD
z(28N(zJ1rBi#DpNMG;B2Z+|0(1??n4c!4_%=E_gHI6Fg~Vlf}zGQNVZio2wIa96R&
zKQm$9i2$*`leY$iFE?
zg-~cCq);OzAzPymZPr2yQkEKus8mBKiB`|&);!CL=Qy6@|K{mchf;U<{k^Z{yw3By
z1ocjZ__)`uH1j;twCD!w)CQSS9yrrkUn>D9ChjBe{1rhK`{LNvL%60wdIoIg#jPl~e{y
z{5|~@zy;eM@4kGQ$&)D0@YbI&0Y|$Rl=^ttcY`u(8o!}pP{dYq`t+?Bm;+yOw0{ti
zo%x)p1XY<2`^}rPy<}iV(gd|^FTq_@5b15-JY8vj+cZ7*is`;D;r2+#Or?Ry+LEyGU`xVq&gJ2&^<
zXsa1+*UrDDq^7F=3)b7vxX14M+%GcTPml2{PMqT~9`H}gqnc2#+a2FF9h^Eyr};tg
zXV12Af_YQNPIyEjMT#j{cD29Pu41S2mwKDr83%N{>vPcj)yDl19c>4A!}6{s-*fd(
zB?fI`hppu+Lc#dl+0|85BKI@gu#Xk75vZu@ZidiPk~+pVnkl|{Dq**Vj)-a?Msr#;*DoI``7cfG-iW`C)%eI=HKQS2i=_tIXoI*)kre4b!$cU!n@wI*(54Pn9d3p>*$Loej&}3O9
zC9CiAvCThUdfDwnn1G*A@XnW!a~nA_4!1jk@ctR?7RevQxDQTBqh{W(_Jn0Krw?vo
zl{Gc$TgEr1rlhE->ZvOTJ{V`u`jT`o+`$uOSSHZr0CDl)u&_hbZ#hlmJ|H9{@fP=5
z&4^2ke>59ctwLzhFXgcR?SQ_VvJYd*b+bpnMIymuT=Hh$22z6m1$P~o;
zq&uxA?)JPTCW`3gARX
zmnaO_1Ta@vKr6s5f!c$}BO16YB&DLtJqG8JlMxVl`upE#DPUB>ba5+N`cNa?)-H=*PH`j>En9ZPw`}D0&Y2<#f@~6iUrtfc
z?^Hh2!Ia5-4+acWDKvBt1JxVGjUOM!I01&jse#(8a!^;_Bc5t|fdONCTiaCXe?vos
z;lr`7!Tshw=5{$b6Mm6=g^`qx&-KXF8Sysfmp(B#;j(@`WSdz4o4vj7A4_IFGv>Mc
z#`m)b7#*9=Tp}n?4Fs=7NW9=)s(Mj~0HG1p!h^edfsH0;2NH&t*>Af_@hTN$dULkD
zy)iCrDk^KKc?vH|6Oab|;UcU-+SvOJ1?Ytj2A|X?O`0SU+1tnY4r`jaP~I%gNMYD8
zKv7-furTrx;xR8kMUOk&Hb&T>dz_6D_nSR4LtF`R!t3eEj0D5QI}Rz#Qv|i!+sGMf
zV}p^F@b#y(FtPpi&prS(Bk2PpXYBhaUKozRw@M0yQoH|eypph#bz3!&9{Fk@6
zk|{6Z4822x2M(WHh|1$pQ~o}?gFBCFOCN0E@0*1A0cN=j1jDiq<~r;S3WB$0$I#RQTyN|}ijP--i9;$Fa
zxu-qZ@t5Yfaof5(r^-Bi_6#e>1hT01IJ%Jo!4u-b7s*FOMlM;td~P4cC_0}l2Hm%u
zD3oErR6il*kuzM1;%jSNhDp8~$6ie88gXnLVtt+9Jj=*rmWeKKK%`}kebl%@3nEZI
zm58^F$EOynWKbt!E;=fpdGJwr@jiqGr+d-!2L5$@nlA}T)-H=^p?9R!Lt0pZbC^b4uk~$B&Ex=Qh;|P8(^kz#fBIY8w%Pl)QnbU&uf1NrZ=~_@U|5KC
z^b)DpXw&+=)tGPbq44Iu%tHm=yIy~PVqii;n4j9&KSR8y_<1U|F#z07r=!F>G=ZTk
z?7KRdwN0~nZ4?*^?$>VX=h0LRmY0{leXYVTEI4@nM0N%r@|}3v3q6iV^kHpW&H)~m
zQ{3Kd5WJST1_#xzT8yNIHw-fpvxA%x{es9%7aHI-N1qQRr;YyDq$1S%=l{GR+cCK7
z&7RhJt4-8cbVBjwgtfcQm!n+e4CVg&0x=k#>uhbR+1>ehfvO&+Gc`^vUC(9}miezI
z^{A;CQLXm=EaLw
zuPlx87FiVUlqsgbkU4nzw0eybDO}`8$Azzi_3
zLzI+ez<*V>wYP^a-Uu4B1x*-V8?U27gsSmO=O=LYY*|%&{=C*@adAcKuPNCXhs`G?*3LkhB2R6W*tR>>0R~n!M6#jeMTgXf+sEeuJ-X
zdhIF9LC)!m{nT@xzub8>URqv$)H2D!#VWa0CvyKa@)DTp`SE?HvjK%~!G;^Y%mw3%98Lb3U@tl0KU7{Eu-FdvXNnVyoQ_@wmso{QUE1^#s-nFV>3vt?XX8
z1RX~=XTI>{Y{wInIkfxaOj!@jA1`M>{ZgW&Uam2_Yw&ujsR-@_9MceG{U?HReXgHlvPk+57jY>mW^8GZ
zn^@7wAc5~>^NZWswYpHIX#VJn$4ox=XdUx!aJ!ZV-0kI!1H&gbSmm%09{&H4uNIOJ
zJpk)0V$cXWo*Df2#*F%{}loXs!EpywgzvI7B^Xg!Uw4k6M>iU&dmpLha3dJN3D7x>$qqf>h(M~L1{ue($
z)5CQJ$k^3-s|cFTB75OKP=JO`>Ki+>=eDlfb^o1AP(|2784GjECs1%0gB(2q9IP9d
zvq(IO^*zQO$;YX`(DnrvfJuZz^~xU+glpbxQI1YiyueIj1ynN$5+v$i*U|mNdM=GB
zNOxNOB1W0ckB;TNYZ4t~Y9W+X8@)8N);=)BROdo_#I=5q`dqt`djD;MbQI+Qt$2U
z-4fg!BA5t;u2LVjX{U*+t1xuPGRjYaE=HfG7Ql!EN{>bW^?pvTpK|j}2GGcbbuThC
zHT7GMkb-PhL4^%N3dq}&ZEq_1!Vn6xTsXogHaUcneajEM(}qy)hyw}2^BWG8W|~?cS*oJwXEC%
z^)KYDq`aBH@5?{CZ^NrtTTrhb)vBo7sWAGc!&Rl1shMC-%vws+Kq1yYOZcP6W32#7
zi|tTQQ7@&FgG&*0emB<5?RlFbt3-jA66mwpVOx5}p>jI88|z=_s16-5f)MyH>u9vg
zgh`WZF1$vX{dVim@DI|L)2!jkiDSx3>Tm0%OX0n8qif)`hGb9C*FQAXzOOQEdx<>-
z^|f_B&5QCfKNd4b(>iXDq0v9-OI-JMYg5y~uVvTwMk#BCvi@=HjMAGP@Aj-L&r3>4
z(Z#;cRucD^Y}XR?loxUxw=^YOW$dKQo2=z0vGyi6CSgo|X{ir9LF4Xkzj}By0b1rB
z@Ap@()}&j*tI2#-RYhtHXz?=M{=LJ#eIB4yN-lG=e}2FDh7IF%&24U&P>?4l&adGv
zf(nVNj~=bN*6(J$9qzA%oItA-p~P~PC8u8niU-}5
z`1$8;o!?SmxHkdXE0^RxSh|UZDAX0Ej8>$F=&^+>i)OpbwLuQsGJQ
zo1-#hr2n@QS^F&30;+r4
z6*=ZdYBo>qwW|L=p?X%^Ts9>}iJlqMjm!G{XjPud
zAW8+PwGK`R*#lY8+}u1p!=K7a=u>5|WCCF-R>RQ~U`xfQ8545-A(|Gqvq~AMC3|dA
zz#d?Lh8+gt!CyMvwR1|zF3in*fKyu@z>;L{=+tO9ZCVGT(-w^Eu2H@kdTHdMu&o_0
z-~e@Xn(k^v-am#c$cL2GOL@}GRu?H!esQR0rOq177h8OMCj2Hz&qTMdIPcdsAfpC~
z9H)EW#P5NT>vTxvzUPihmw5
zR^ZT=EHLAo&(U5fn7+h#Y2-n<503jxF3f)^?xooSUnfaA1YGRAAz`N=vn4~=_=mW)
z>SJtpK{Y;M<5pfj(Dt0ajby7;H+!eatqXnT=qe@6z5o6*cK=ZTFu(Zu_-L=nh*V-~
zs}ZN!rexeQh_RwmPZ
z76hKX>0&$K7O@~FiEJg?L=VZD^u6M^=ucQtiqpV`4HsKh@8Yub+4`kwBOt#IJjZ*j
zBoP7$C792hH}8yPbTPGYbuxUCH*eM?q@IM3O52vOv|N|Ca+=*=Y9P)IzijF&{SO#3
zZh}TBe1XXG3@E0*b(~w*S3xv;zfv}hMvD8Y8(Y)q>S174pm+|6t?&XP7U;7h?D)~nKle8v(l
z#akM*11%4ON^5d3GzzHBuv1BW@M3h1&@6RO${2s|yv#s#(L10Re&JEnJ=$t2GlJmP
z!@Lh0tVfqW5IQe!KloS6r%wnZ#*Q7EpO=@u{kuQPa&PZRmv6&oXKFu3A{+vTb3`?V
zk-;WC7Qk*A5dyO0ix+6=oT&KO5eqwF`ErHJnt&2Kkq6C2v&_FzX5+UqNM61ukRD;<
z=FKCa;QjfhJAen1McBObcbHwYp)++>1EqKRSJ=X!260r?B2mSNFUOdJFx*Ey
z(pPF49gd8&#hdy21{QWB#Le$^mE%P33YxUZu5kC*0i6S`IBn(<+04E8TmPE!&967O
zx>~<%5PjP^qbW{3G(k^KPu}HR$)%JOR90?{)=AUcempCe?HmN)Ju0cb3W91&zmFyq
zd%=$tX--_d*4@0kI^rucgCkwn?^Ew_jZa`1E?^-hS`LJ@9?h-EXU__gIlz&+`!OVS5Qn2}jyndW|a5P;8jpX6gj|c7O8M2rtO3Sf6!D!7bvGnCo
zkCEY=FqfY<5v5dRrHrDY(5&~B9Q?pW;|{V-pi+*6pfP?E_zU|2sz1hj8v&VLwR0Bc
ziJW)*at3d)+9l4_f4%vA8$er2i`SFv)S(lR<(*l8&CXM%{=?ki_cKy}=um#eRW9i6
z!R+z1mLNgUsJ`8cvkXlD14gtfLfY+
z|NiUrjc34|@y6u`WM#3Na5zvKa&NGB!nuzRO!MnBBXc*AV(PZXu5)pberyE*!Fl(s
zydw-$!ImQ;WTECk10oz?=k`0{ws7&ND%F3F%AT*?uG~X0$n%aElN7}Sf}|qo!O0bZ
z-VbW%?^8&sX#Mygdn)0B>oM1`ZD-xnU7@Dbam
trw-)Pf0Iq|Kd$Njre6H--(2{;_iqvY8>U5eUKXxpj+vDxdDfbM{{YwtLiYdw

literal 89413
zcmce8bx>CA`YskAN+<|YBHbV@jevA_Bi-FC3Me8V0#ec~-O?(cba$t8m-M+`zi-dk
zbI#0}`TcX&e0w${u->(v=Z@>XuInLKK~5YUjQ|Y^2?6SYZ(hURD>+moA
zFGO#{|J}Bi&~!pVB7TGT-?i|iT@xgv$4HW*&s5w$Z%n!A;QX3F+2NjjD&$#2mWFfZ
z=>CIZE|s*LV8gVZkCd$yjbz@lP+8^&dl-hFDw7io=7?^ji5a30zIr8L*MC;A93C3k
zHC5t<;`Q>>qw$ZCotvwxYgp~X#TJgg4ihze{@=^H*J=gvOo;2-`VD{Nf4)STMZb>z
z@0Z^QZxa0bXgYEA{PA>8oOG6uRRauP`iibb)IX&3j+p}CTypFU&)Wb+17Dhyy
zQ&u(Y88V}Et*St~imvV!Cnu+2$=6~0)6-Kr<3CF=`gp=wWW%@Db%UnJ6Ijq+uJKGErn+p9UZOrI`00=Juop*>$W}p
z29xAd#}c}(C2XPa$ZZ8uLBXqyv9gMZwSMedVCGhN$5)!>GV@OB@XxGtGv@|p{HeLC=|2%1=M69Cu=Z4cYdOyFb*w}~B
z+URiIRXJ7;d-rawH?=8_&CFSo2k9kpdJ_x5!Qo+7XQyRJdTuW5cHmrMVxq2|-lH-d
z4GnVxgTCUWxv3JPUthj_5haVL#U6`LYG%^xFrKO{*fme^FS3lYqYqD)o(9;tzY<|15
zvy+mNa>~Sy^fGMqNpEkabj-?%`P337kE)Ii5k5XX5mEk--U?S=>jNAdduokp$EE&E
z={S=cH*Qp{!ply}kocYN_6S~e>S1GH?F`5Zo}8S}W|S8f?;Im8{ZeB|clSS&)lSOF
z%5k!)s;byjGH-)}9o*cCvK?81e0lYPYbWpd_HEb3;P`PgL|oVP%O&x(g|VXE
ze08`!BL4h2tvEhTKtMoqbMwJU-|IqJ1_tk|i(?m61z5>fLO0Ie6crUkMq-zF&o>9Z
zdxuuwnteg_{{8z=!8EuVQqh~q$w^w}tlk)irN+zCb$A|*$*J-2)6O_H2(~OMCjAEQ
ztx4`A>Dx%XDOlz&uX;yDNZ3s8mg#6|5!}0%deX4ri!Jg7gM*7}xzQVOPxQ6*qQ>sF
zwp3<3W<5TKdC>w#^ZpD;4UO1_1_9kN(=LMPb`+#uR7kRSKf{QnV=C78zdu6$C`OO?
zZ9T%kz(B8~jZ|h{=apjXu~MzYff|?f-d=^%z4McOSVX=kOr#VCcN5P(bv3mlJ`Yw_
z*1gM%vz}x@W?I^aj-|rCm#wWWJpUs$)Aw)QyixP`+1woP=1sDop8!8UKQHeDME>w_
zrE<0$#O!T!^d+CY5Lm^9U*Y0nVp$m(#wI3`nMx=qD0p}=4L)bcw{E#?j#n^oUmZ=@
zIXF0YdU}%h{xK-ht6x*DB2ab8W^3|5FA3>PKihF7<1yuiiAj{NkA
z3O8hGYU<+rJWD=#8uD|TxXPd>$G}2EcsftQ%>@GqNeFeU!N*He6W^yA)-_c@qc(rq
z@9F}Qnze&9ZsSMbopkZ=VS9aaKQkA%-L@hqFg7y{{5(87l$4n%Dc`<-|2{P2(wp)O
ze(dP*aA?XwT1pCom^`T?^z`)T;jP<&@0hm2bG(kX-~%syh7wv>Sa^7N;Ns$T
zC2(P)qW+nEjp^;}&EEEmDl(DZi<69O{`7D?H#Zk{7-I3$r%!r%dQ-J-B!a%)6O}er
z>YVNztsaN4{sW(3DY~VTSmUfaJ~0jt4_o$_On!BMg^gc~i;ZpNkNglA*fL&WWx|T@
zRW##wrE7z4XlNLP851ARWANjqztFc5qZcn`Hz%s9L-^qz0=eYKA|T%1@Ndk2V$^2O
z8}YLlD-GP5hNZ365wM%)tNH!S|3!!>VHC~pMmk%!5!)EfjZp_kD{|FBiCG$9*a(lo!N7-wCaq3b-_4KJm
zq)fy4o^*{g)CZjfq1PG=kGO0)_#{Hbk*aJ*z5boSe!BBDop
z?kOf+v`VP#=YJN$Et#)OyW%NhI<79y?4TEKMUh3U3>RnNKtZC6_=OFqo0sQs|^*X>^k+HL(R>iq@j_K
z%Fv%6XQt{roR)}uPnKfh(TaAA_S>kFX&~lPo-xG?Z2Lq=sAy>9aAKcE8ygz_5EbFL
z%`QQ4|LE#&S1MGri?d_9p)b*UcHf|Nd*qWGRkH;MrzR)sUtL~6Eg8L{YiMYI=Q}Ww
z^$$W8Oy~Jxl=3X{{+{;pnpQ*L43$7XObk-&WxXHjl3xtS+R9c;^E(;9}
z{h=l(D9CrslaXUhNJLH!<1D0qa`HUEuA!}`$1Izck#VZVB~O?w1JF}$=M!Gugeg(@
z$k`C)hStu`jN5ed^dm+3MOc+}j!S4$7n9N(v=73FdBsM`ChkQ?MbS8itWDKc36DJz
z^yTx|Gj=5)Cr|$sPLjtUcoy=QmGz@9{q39A_4Q|NV^eh|@*e;0N*Et5YZq#Q%DUPP
z<;iLJ+ZGp#Y3D=JOe%lmKxQd3vt{__nCR%fn^5tReU9ROR82q&anDVL`-X^;)r3e;
zoNm=_0|+yJ1syV9r}|@-#n6`rvO3BssZigy*6y4476+WGMKH16L_$)q0DN}f*4*Cy
zZlMiVzx*O`=>oAQ$ar{oH5+qtCB7G@fZL$3Qc+S8b32MwuQoI``svbrCN@two+BkC
zl}{Ehxv!$G4yCj%s}MaaD+@qdraIo(*ch8>r-DSfcf(y@W)=5FLBp%?z9ZJKurR2&
z={*E-sgyWDD#fLxhFlk>i_6Q}8XCRqP?@0{FwxN&W(U82f3UxAWLDvK|iyW75^;cTU%Rj?x%3U%cqo-&rLkhbMLZGHuxkI60sVi
zx?uq9fY&xv`-09+Gr8=x+8^i>*@OGwLCa-#X-SFV7>mrr>Fv9BGgc8mvc@PpEKJRp
zHHGzW@o;mw&2Fv)UG(koUBGIV_wU~q_7gw9xLBWiJAT={=ewI*D|CaycJgT!Oa803
zZ&3r^_yIb~8^1mnKy3a3Q9s(-SlQU3sm+g%j?}ncA)j53#d6%4eYV72r=PRq>(_iq
zk{+n?R_6)DeMuo9BwiC|=W^kAfC2plP-pg*I_NOtqobErRwR_e2splEmAzDt!MK4A
zWu#I|tIC#`m+PpvHj+wqUVy!paV&~fISQMCnz~F|^5T5Ihp2jUwB&Wx;K)c%TbrUp
z3TLl^mR9>^5*ixXeAo{G_zl=;#H2+xJ!F+X)o
zxuA+?w-CNjZg~S;IZGz>_UF%^9i5!GF?N@e{Yw2X9$czE)EdI{B0~G|E<8sRB``|`
znSkKvy}M$J&}%Bn#rk*dJy0uBs5XLgU_#Nv6dzo3Q^UhcuEAix`-m?#tU6}{;PA%s
zIWdp3g7(XJnwtbva-TJnrH(4FNJZu4i?YlVGcz-fj*i$f+gopKA`|Tr!&|bezkmPqA={zrC+8uLkk~6%QB?fiJvUZn
z=IpUE`+f3_Q4dN}lIz>?2A^6PDx*(6K0eVCyd|P7~8F)UEwwG^;P%`hiqr6kREdM6mD(Yx1xujPe#TvZFi=hry2kxkMfZ
zlEUvjIV`hGq}!@0n17*Lz5DJfCM(uiK~iHF(p(AhzI>VQ4n1?Le8ePhVq$GwXEVXM
zMB1MN1vFgX^ld{OWCsM$ny-$N(>`>y!E)8P)Qk7{Y&cmp8L*o_8m`eYGpkFagxq=H
zyB!`?pjGj1!Ea(qRug$OLP)16i`p8#j&?
z9HIy1481o-S!(!9+Ht~|IfUQBzyvS8+5wr2-hM`6*8MrP1A2iP_W?#iNy%=8WaLUm
znOToxHeRsk{&LSqrHYova-!RcBTRm>Z5Sj1GBFm`!Ud7k^8MRXP_T|lEK6T}zh2|L
z@Dsg6qccLo=7NadGm6PcJh8PS*!$}M`yYIcrxl0*bJkskalo3AlG0hzzKzM`?Lrs2=x6bgy=EJQ=%Z>aSz^cNvyYPNoD_NAlLQoecxIIdJbjEMUqIelO=
znzKXdkL`x6`aOBS*?dlMyj4Ht)`XH67tP1RS4XRFsG58fo%EJ
zo@!{xR9!ct-rP~^EjAl`n8OHhlCjQ%f`s(K=&JJpF7EJt-_T|E+Ub0xTYLHqnzD)t
z$SfKkkNw|WH6-c=6vDUAZip0{87TaGG%Op(#H8Y0C#W=>El+av_CB3MGHY#p6AVf6
zpP#0xUYm7EYZ4M+M1FlfAHr*^
zibE3f*~`R&HzD%A0P2egkDOkFPj_{s+sVtzqbtY9#W|GwDCAvZR$;G(_XIXzoH)!d
zIZK5O9RtG`KxUC{t(aZ&Y*(tG+@P2oQ;eChF*JAzXoms<0-T&jN1GFtjq3vg1FNjO
z{aNJI)z!|LkFfx=Tb1Im{bI23_}(pwUn`%u^`4m5^<;0U6GD14U$YG7{$SzE%@aEm
zHh-a%Ihu!nshRK|JaEW>S1T`PugRaQ6+zuSJT!6RQ~tlV5`_S!RrWCxs3L@%*0U0`
zi;ED~nUZRgjb(@SyKT$>c4U)wBClKzmoozh9;xR=(G5)mN^Ne1M?t%`s114Le**>i
z=1rXj?>JnBzgK5>x9zu=*Zlup@%mm(^qWJd>`^f>$uM*_2V-;DO?&Sx2Jd|(w*;!9
z*90A9JNM2J%EaKHih-UEhMJlh9h>6sjSZdJQj1L0JT;r~a;T-(NEh3q>0zis`(<*R
zpuqbW9kBA4nR;z1%IF0}w8IGI*_FTU^5jz`PDokgNkq!7rFKI%9UX0S#v~EQ&&lbn
zV>FY6xxvc8+2rrr9c=_n5)l!tIKm|8;!+WG!D+bq`AL?W-(~QjwKA$8HPq+hX851ZH(R{
zB+T1WHh%Hq8!@5Ln;RY{s18Pcv^ksnEkCy&{3QONlC+>MZoYjxPUh%ryRnR`xR*Y
z$9Y8o1?3HVRRh-+N}yPdA_@U4)y2ida!+#6c_0}U>g3cE3g#m|yXiVWlx8L-1%MRE
z>5slr9q!C@Hjod|HMAzpmTj>(>e+YlE3W%u*X49DlAT>WGf+~+h>nh~5J;o4AIQVq
zRt-QDlCG1pvoif*ALN^^1E)Ne6nM#w)jCa48yMFoaZE3)Db!mJMNa;Xf{C6m)Br67
z)q({9ij&DJVaQzSh=DcUov!yf)ZxoP39VTdm5^u!7HU0O#RzcGNj5)YsQXOG^txhzTcDvbUN~*w_f!
z%|HD1h)+mZUtgEbRQhdVXqYyRAN!|eOCa5Y{#n@7XBbz0h9A=jC*D93;&*a(KG@rn
zl8|_zw*})q;LLb8C54|!A4#FJ~oR{purgynK0H{S^4{Ny4b;@|D<26
zC|NbkBEal1>DJ&76Bq2PY|S;jd5w(zE;O{D<7+}MD#tV~`f_)ok>3eWKrO$1UG4rN
zfj+lAlowN(Oy>HSk(M^NN@@7p@bD+5jwQ#_*UhqY(Nt7a{V^Q3uGUCw1|P?CSlG`&
z<#l$QoSl8qbta1?eDnIXvls>#RkA-?zPmu?_t+C{s!tN|(HhA5@AtbP1e#5^yR{Y2
zrr_Gg38-;vI%$6FA47W1&%M!=OT#-tT?Z*cSNAj+b#-!fy!=$`?%ut7{rYuaS@Z7T
zF#(AKclCr(YU8K{CfN0nBD-(*Uoo*y3pGK1TyY#J)NN{Ns$R~(327nH9#fu$(ct&e
zQZA#Asp;>vm`a<8tY;BdCRSD*kMK2`n7qRu@f&ivEF>*d36#`6@14eW+2HE?qDo(L
zzBAXv6IIAL8hH#8x5a%QW=6)P%jedkY@U=mde3t}7RWIJoM0L*`qjKIO$+mf+fw@j
zo~M3O96=#H85tB^XK_>%6y~G;EQm(QmDFJ*yW@-K*@~~Rf<oCFRKI4W~`3*puZ
zs}X`?m6SW7T)jIhedINKHskEBz0#SGI8N3!eQCFNd%gA+B`a>BVq)r_-#^RJ5`iHx
zy+b^VsLwcNXUD!~62|KztXlLuXrIn$3os6`98Z*6ysoZ30ivg(E0X_nf3?#x``B})
z82zvIbLZ0Kv|5bLE4}`l-?2hkvgHz$YNbU*Z@N`A1@Wh=U;ih=UV8psTx@>SK*z%3
z(DsqCqE(y$msz9uWuxQ0w6hTT3bz6RxtZ1vZN>m3)e~gMB8pMSx2NlCf4?!eM@$V~
ze$BNo(iD1IEc^_e3{K#(yPR^HDLEc;SWRi*m=+STw4~EL+TB%5(6gB+gP0mvg5H$V
z^@NP<6;qYPfS8SESQC(t85tS&*+u#J=Z7Qus~s>NT6l>j8dhmObzZ@$ooNQ{so=;;
z4*%>X?B|y@<3S(=H(^Kypm%n>EePemrqkNm+LjvEDYJ$0z6NuU|FxBShz-d);rKZ`?B(@>L8947|8F
zo|&{j5Lz);r|U%kJ{>6<{H`to@8GR;Kt^qKMh6B3^@k}94GlT{?wpfrbX<};vM80F
zcnF>Qlbky?F0R_~3DxP*=K9i73xm44dQSpZIB&fQoo<6Sx6cVMiX-zW)*A!@KBw%w
zg&7EP69}@U?qVqZy9@t$UGy?v+v!24=7O;D6*@WCenY}(U%qO0RwiZW*y
z{w!x>u`N=)NN*_ZYk*Fzs|{>0(QTOZM>hq4?u6tl=Min1zYq>QJ|p)nsTIZ&CL<@W
z9w;-i>F(~XlPSC%D!zcuUi`~s>5=&|$Oq@Zla5-cbW0FDd^nmaf^zBz@QcGo7uq(=
z8cE%+R=2jgFEX>9tYS1xP9j+TgQdQ7ac-^Q&8gZL{tFqHKE1H~f2M#6R9>KEZOyPr
zYHDG@{o>n8En^xi)X_#i0|Nu;PYk)Mcv@r;?-BR(CdWY&yf_%pB8p<9p?OOz=CfUY
zyfxEU#A>3WGwHQmpUv6y-<9fhEQ}+?22G<@xXgN!kmo?qI&Qlkt_|tcd-l=1&dZ~(
zwX-8$ou3~V8j5Cu<%_=%O=J`h@Le0st)STl?v+`$Cc|+6W>aJgqSi+uyCB)*wvZBF
z!C+W7L$cfP>({TbvABlwxi=(FgXScpr6-CFZuqp>THP(W&0RF=U$r$=J3>kC$`_8c
z2Vox*6GNX-?XjQrqlyofg-<|0kGp50%I;2}aJ&q~eV_druW{uV@ppuPOQT9C2Mo_~`H-RT5eKi290WAjfgm^Kk$O|A#fhIJh)~<6e
zX(zYT7KH9&_(m>)-J;ZXs^-TJ;rJBL9{|>ynwrX7cXM-#cL(Vqbm#rr7J@YvXMgfU
zAwh5A{+{0RIx|$5!Gk}3yuW06e%(I>5*hk$NN8wRM~9IT5H$ksJ1?M?05aTY238Hm
z8>6bb1^M|ZPv2EGmeuKdmwNSK3u{$ax&S-vdOc1Cu@1bvyj0g}`TZOnONDLW*{P(U
zs{r#H6(8STH~|%h=I+D3dM~%5@Y>4JO%B(M(eDu-eMD?6EZPz`Vd%G*2M|e1J$o$t
z9Pe0s{4%_RTncWZHu|O4vZ<$snU7D00k>=b78UaA(Q{Ax_PYNI;J!G9B%6>4`uqE%wAm3G#*wK`X#SyN2{a}%6ajp=8*Cu~tFgqd
zFyd%xmNlRde^fnC4LQ|>O%=W*i^u^URwU@|>Tm&JEs80uJ8*Ea`j>kh%%S`5*Ca;#
zk-2A7GeQvL6T70OCb#3Fgc6&Gh{#dJP3=6jg8qSlCC3vesuJqM{JoEt&*s^vsm14B
z={&}A-W+!UH7x{}5dqT1lyy5}Sr!1zW;0jcK-aBtPVI4wwXUzL12PLbTDp!eAz=da
z8>^Gcb8(Sd1dDZgIVSgqR?LW{ul8Ud$u>>ihZ5j+*0TX1l&qJ+oXk6QpA
zWihggimpHmBxEv-`i~dDQAGllcG0oAu~D%08&ol<+Z>d`6up3yA|oR$wmB*QYewCu
zQrTP+k2H0Mg{EU*sB_;X(=7whL(EIeMG+I0uShi`{qORTMo4hlO7m(s6uZVy4zjzR9Ub
z@&-H{tPz63vqRABAhDMG`lw(93KJ0*&d!=;9b)i!_4>7ZB2NoO6dfT%0Tk5%DKm(=
z_)d@g)$C+F7|y(~ZoPQj>JFE|6_kcP#7#2GUIRUfqAWow>#ppiw
zJ>hBJGaFa+sBut$8)2zGQc8bq{P;OB5kR*Cz?p)EYW5N+;sAz5CnoYHzhq}ux$n+H
zNQ2CdgNHY8$ibiTICS~KVsmj-0purs
zYM-*jh0}|Adok4aQ3Qzh#l^*M-@XO#wY0eSw#AT|ni_r*##P=-E#CJGicw6oiupYN}-n}HF)4JxkM(Ee#_
zTN^hB*rKA%pbVFlm4P@4s#@ab&)^skFz-#dM=oZ5ww@XN@#9DE3&7w5BLWZ&u?Y!W
zHsj7PlmoewZMOow$s+kDu;yPbuG{<NEUh^ZBZU??H0Hxi(U
zQBqJm>43VA7P<(ulgI)A6r`yJA6`;XK(ux4EDQ`AFfzeV0PPF|0|O*;7<&l#JrM)k
zx*ldIias74PmSRG2rL-Tc-|1)U%!4$OG_&)Wi>J~x^o8#kd`w<=1OlWY_7@i#>enC`5_`S%B3s^itjeIv94lGphJBrpJDOA_9{JL9001Zxft3>e8TAdF0K
zS);PU4+3=y&9Spolnfu{^4*1B(Z9qVgtWj~KVmVwY191#MC~yAqAFlzI2ZRx*V5_UE&dl1{(77)v=cN9zAoY%t
z_9EF80^Zz!!5+#ca7da6h!IN2H5@fPJ$(s)2s}QAszQH1^QAf_;)NL5*$;s20H$t9
zjtUD^ER+D@Nq_-%NdEgj&!(qn-iKsY1NxR;qwfWXnJ`T>fJiSaEDRV=RW%wIcep^Q
z1A$0EK_P+DhM*?npWiJ)^50u=`{!#Uq>9&t*#Cb0|J^S2|MX#{MC=_Os;;n^PInu8J^{TPA~F}!*I+0i1(V~LcY0N(>YX>(_%HK>Om
z9KfwCR)nvue0U~;_s_3{Y_S`d$Lz2Ez$Lj;kXl
zZ5|wqTEM7=Qo9iY?>+U^f!v1{H>60X8Zj!vZ-)_b$r;Q*L!gxYw6eZl=H_8TK}E%i
z&LeQX)ASb-Lc#AHXkvTe?(?s~PJslFVn8!?Xp27;2z!O}Yky@6iT6sWId(I8pi+HK1D=2qQziO=dF-gKe|6
z-hd7Y?j^*G1_jv5-Mx4K9+s7!-U@mmOl`l{y$QqH+oepugS1-T08A8gGDfZP;))8J
zw~qB+2+q@QT0*8GE(H?6n-UTdH*jYmY#=zu0?I9iEMa6Oi+F2$KUfq%o0P09zx}Kb
ztSb1rYPTDts|37`V9Li5_V)4N;2Q>%ib@KC>u9OTow6*b(V(V)@Y4ZPH=^GdwxGFQ
z0x1k~t5$=zM`F*OD#D*;#Gg`)d_`lCnUk{ulPSb6nC={0U2Rw92eHe006-!k&9Wi{
z==`q`GPKs?WoDSfQB1V7k_PKlGrk?xXrT@$r>*Gl+q#9c4-Am%heVds5zZAs?X9JAXhe8(WX_#TH%2{$*J5r#q21
z-~(bF8VM0igL;WAdflehIf8ELeXAKV8V6euVt%E|*ET*-i@
zZ14-V3G6Ar-m@X>Wu0{`AnA1yfal0x=!tPF){;?XRu+?1d23Hk&yOEJz-`deV+X4Y
z>t;!hFr~zE+Gv4o2*xc#D=UB$RXqY{GE0K0jNNZ#C&q^>t%+3RTGKYG%`dOzKaXa;
zcnuk#U2e`|JNXsNJ-}|=6Go69M9)pA34VfkE0!Iwa|*q*v^1>cV3EEcGcz+bHa0JB
zEtKJg{cc|9oC9CSFC750z*4zwO^!i|baxj5?ty9%5*|LnNy6la*hZwBF<99+`eo+E
zb0vcT5Per%4&h%OR35F5gpWaMW;B(B;Te?_*0!?)9w(C$e?)-D%w_x1RU>M#JwcLXq20Y=H{C?m^GqSSiGUR1stOv3}+YiVh%pqjk-@KTP
z0OI$-g9qqDPd`X*%+3mlYXD&fgJ|Ze6@lYCfE$_XS!u|akMiwPpw`C5#?ow?TUil;
zPLERydx_?ys-hAQ6!e8Jt)by^MBh*MnC**;%}AOx5e^hfUc3sDZ#62bkWOB}7`lpv
zpFHJr-+AdVma9x1ovo&*Xkcn8hRVXl1?EI6L_3a`)o@}+%rG`6}Y0x*SCU~M{Z
zRPe~jewT4=bKp6F=?1Rc!a~#Rs)`Dew7Wo-X%@fu4qN?+L34g~_S=SuiAi)~VnJK*
z0jParOF$8DWTuKxFD@?jMzfi9Qv`r>6SyQW(4L*0@jiX}kdV;J(^D>8*=6n)Ce8xC0mp$)##CtO_F
zb|ke=pb5$tpprsw05qos>K(ME`uh4+uNrU%2BtsRmVmBrZ(j`3Eue$_y}d<84^7Rn
z;~772iOQpbd;u8LxPKXBRq#s&2D0G$?zK^OYJ=Caok9WM{Hgm+iqPX1t|gU~V7gs_
z@PVe&bGOO@-rCG8+Z&TMTFQb!k_HC@lpBC_FlYlaPtTu@DR^ESW~Nd*UZ^|v)cz&`
zyWqbY3+gDy9JfgyJ$eKX8@#3-uq}}(;9Tx(faatuz`=9g3?(;?1MJ6H*Y03qhJ#Vq
z7fJ(^FcNliEGMPT#!tcmbT`nUE*kVF#K%JpAdFM+eKWL9@1N_JlE(kGn-pXxHU=Ct
zK=CW>@8943y*Am~`vGen-gbHkEE%1{(A@f>>9)fl=i{OfoW)b7`fD$?%vLVKb(omx
z=+cPsp{THPhXBLIwhG3-4AyIGaulzp+*5%k3cvzmmNMr4hYugXUErz)ZqLqwh~(rM
zAkyKEBl(&}c6Rcpy9Wmu!O%h=>3E!TsFs#-A3kiwF$MoLtfI3I00-b@V`YI+CD$b(
zBI?YC?~tGQ%2P%5B`BpTJjBEMC4jac22>RO!-p`Iilp8Jn-PBGGPsOzEMIFjS`N_#
z>PFo_2Zrll%*Ec`UQzMh6gX3dA!oEU(T+cq)B3{9zdcp^Yub6f`8GiukJE3NsUC@m
z|7=e=KVBH-3*9eKyTHCmQ``@DU`5p-1bM;?z&gYA<+4LU`Tzoyu=6^gq5*?^y*jmPELA1w%+#>RWh%A-n3Nu_Bw
zp8O^>r93Na81_ewF34ZKiKXwo8%SkV0@#(x2MA0Xb6QD6Lew&-|*SOls%ih!@RwKG5)Nh*M4+{}y`2Mxx(e904ToodIx
zwnhLj?b<+5px?cVO8V^Cv(9+HVOec2D=Ik7;>r~BPprKG;y-O9H)$sq=^5nsmtGx
zFZXe?JVhF=mjat?M(K{Tc{Efj8f!XQS_3WNlRtlYqdT3;Kso~9uDzONAqh)(Z-0Mm
zJ^BVZOx$^^dB7dA$f^Jzp!!Ku`{CXDh=w5)3Z{Nmka~1d=;Ju=fQbNMS{W`D4>A4(
zuZDd0p~CTX2JK224@zRg;xAvMB4va{MB-K-Fas{~yY%K+CLfp@y|hue&gIIn_&Y&<*ET>m{i
zz)x)*9pnMtr>og$A{4AZ@pl|t|LU@iTf2&ud4g}DJ>kNfQ0eaBwx(xIP^dIJH`fOo
zg0O_MGrP^h7KE%uM@QG{hoW?`mRAHB+LOqe?RZ!H&Ye4^{=TMGKRSs;P$i|Ml}DSM
z-+wIgd^dFOU)XM0cQ~xwIaKGELi+geY(dpDsQP62LL$&~;NGMe(Eo@;zFHJQF>j|=
z#lUU--rxU`^TTwp!RxB3V?YhiE$$N#tSv27*v|`cMaoLIL=iNJ@Y<1v_Ln8z`5Id>&BGH-!XZ|^92Fb`Q_O*uzsLI04#0?$0rXT%FUZl`dqfB
zF92My|Me07E}^hNQ^bQf?40#;IR!p7#%f`d5(c!3{AC*Y?sO@)`U3|CT7$tV`fW9zT1T-V^V**)AEEE$DTP+s6)
z3j5Wazy-xW@XZ^)gFf-jcA!TANMH#AsM=Uu%%~edblK_p1OO^gQ3pUHfX*8O=MY>u
z1F&F;V&nx*7JrfWfyUc-4g_@L`TXkY7?4pg6k$Gm@@-&1c|U(pIOLy`Qirh?IL55_
z*s@=NcO77uoS8|k^Zp(8@#7D^i>!Mnde_m3s^}|rU|SG&cW}-@->Zj41Fw{wy$t$bN7)%en$!?S!cT${8<|Ra05uB!Kzy`!k0^)h>ie
zOG7gXvMhWTh_)fq*2&*P>}r{%AZ`f1g;PMG*)-{)yl8jsJm9o&f3155dL^2;I`AM-
zJ>>}DF#Sn(Pu9@ioM%X={Nw4ireU3xpd<@>8hecc2EUI92_5--o(D7o7yeUruMq4a
z_^j6#76$4(5G@-7MO!$J0cT8rA^|N<>lP{}Oh-ViS6L3d^8gdlSD4u#?d
zl@|@eYZ#3G93~qiq2z^9c9=lu&=$3CN%Lcz9+u
zHV8%&hOs+0Ux6~{)R*=g4g-LHlN+2(Zb=X^uqtqVf&e#`yB4MlxZ<^BdXe6!sVa&)
ze?tu{JVT3m&0PM0T0Y4bb2|6q_{~X6^@&4e&PPKfqwNb;G!MCKC&7J_sk91l7yFiQ
z!0yM-pJJ)vJ;1YlHR-@Ltp*eT-0(hJ;N1jWb7vto<_2+!{}?n(bvX9|H_ge(aaj1t
zB7(Hj|2b7)3i`N*#uR+FW)E5`OxusKdd*+4^1LKs^QWu=4>=q$+W~lQYT88+0P6P>
zWp_C2Ghn&~j5cr@t6&cW4p&F#Gq}*ilYs7oV+(N71ZcOoJ)vuCg#x4jw{3b|S79+b
zXV2Ts*YEOli0}gs@E}+eN;+sa1+=oRuADZ1g!I2|L#5tc2@|)w20hk*CEd%L58UPr2
z8O9!HpooJ-z%+vjPc0LxaE47Dkg^W`ZzQCbL9ySygW`f%+U0KIz0#SRq|Ln`$oC`#
ze1L_%1dt|kb@CI}TtZGR38YxCsa+m7UggRZ0~cj#Fqot0yaWy;7-Q#FS3RH<$2*#e
z{lG@OX^CP^LqnrcX&slEN=hvJv5qLduc=8SmF-;HuD+!iuTa0SJCTviK3$Z|;JQ#d
zgrUU?#PuOQUP>=)P8>D(?OTO70YSlJ?i_$lUbynQ*l2+;mcTY-W%b{qe~?3T9v+6{
z4oXoxtDwi;B8(mD(|me>GyjXV>QAQcnF_y4ob_)$aS_4HpKt3G%5E?W08ehp%=yl&zpxJ
z?>WW>fhh!DI04Jh+*~4glY0d)zta)JtRB|}
zBR&**?MmaCn!1aliOwuAt9&7n+lKskQL_SE#z;MWwHo9}Y#`8NGs?t4SFj|lG
z)T|P@cMgC1CM26B{i36-ZLT2y3-D`Mxx(N8g3{OXNS>6R-v=DgGL6AS4X5w;&97i&
zi`{t!Zb;a&2e!hX4)9qGGYG#Gdi>{ig5B3USa^8H&}N=Ked@j2g12$V?|FcQdJ`Of
z&PTF8VGon}y~I#ghw|{;y2-`Ej7{0=p!);aF=`M6q7Z0urY0u$gvG)ib>&;bu?TD#
z0C|K@UI@W?L*6QgZeSBJ1F%vRzJif>OJv^`stpLTe=`A~V?426TUQtHlNU&d=>A;*
zyFAF>g@gbpf=@!yGadiw6PzY`Vw&kLj_
zZjAjwJR6qq%2Q@EY=is4pcessA}-ikj-o8Pl2QPG6)&V^Ll&5iP8zn1pf=}UVj5
z0%aXxa7jsd3`cALlO#QLcvY>7&+SOzWswapbvTp7;_7=vq9md&7%JW
zfdUFZdtwR5tN3CeB22gp}QN@mT{j4TgJIHB;>pb0rx^gcfR>f9WQVKXXdhq)@JkdDAk
zu%&y!A*env85c6JQ^M8Ns|e#GqDYNOovC37&o3-QyO||**4EUNs>26rX=#-*_kpU-
z-LHJWeg5j!9d^M$5GkSl3ojLX2dD-PQq?WQ@fkP(9_@usNN6|tRdLfHE{+`LsK?P#
z(N!Qd@=K*$5gRuwwBXR=C8;jZ}su
z?(k550oK}T?n;Brix>8w7{ZL{2A~8q)lX8HpnXCX&VvlXE?5DP4xfjBh~Ucrk`yeC
zaL@@3suOcq;tb&HJp)!5C>7|t7uztd0?`X~i;acFVmN;c(pB($M+D3~U6@_9(A$8&
zb_0tK*pCdUF6~`hh#}Sh5a{UW#G0+PlN*9Snh5v+{9lvsD9}Sn%gVaCy0j~;#~@Q-
zfgCysmG1nlUb6I)h+OajAr?YTrigOM7yCV+Cje@jK%3Trk^(kbkb(i91EFiy8T(x(
z*bxXtD5V*hnSf?r)v^qJbyx_xi(d+NM$F=XfU~m`;^L}nYrQ-?V8my9@}xK{TLh$g
zQx*BY+xAjj3s97)X_f-b!Oo8PO6ksm3s3sUsty77fBkudSQj{(`S<#t{q+Q7bY1wZ
zm&mTQ8Rh$%8m8Qv1x!q2#B%!jcS7snFHayLq5YNga`G&8<=u=;**fk_>iF(h52v*B
z6)5F8>$`59y!ug_X~Nw6(J^jNjppD9>zbL#JxjY~6T^H?6mHUGG_+&WrAb;w>4N-z
zY%IKpoXFf&uQ1n(0)|`Tc&is*sjoV^ElepqhwGgfyCuHopoEA2uJ3)~VDe3yjK3W7
z7824Q%)f6cq&&e^hDS+-Cqc0hjjuWc|M@+5;CII7Qdd7E&zPW1FzD_dp{aY}I~nP8
zD{v$F`NutRlO?$Qs+~#oNL6NJDsBNMoiS_I!u5D}EyeFfw(x#-kDF&f+z}$bgB_Xu
zOHlPKYjNA6vA(PAFYOZAXjM9FWA%{QsfKJUlNY?|3Zt7x^`9+EerNK;Myk_ATP7tX
zeE!hfF4X{kNdU>6E%iAsO3AmatOZ
z7vV<&VU{y$@~`D^ZZAD>N}}3zkB2o_Qo=7gPYsDItEqf|{7TFD5T3xj>n~DMQ13|K
zNL*O!WSBtJZ<Ac_
zX*VucDue23SYEb{*ZGX$mAjb{Z!xlDe@;w+tjKxF$87gLkx@JXjpAcs
z+!$d$>pRLi4ztGrd%Mwhkro>U~=C7XZ#OsPeX7%gaA
zqa^)qR8jEm+^|5RZHr2)alnVJ)i`~27o#dEjfl_f45Mg)XWsQVbh>=o>HI9?2CAO}iNsu)CRoCbUs_;uYU2}F6Sh8mYKguY(3mC1qt;GU
z@Nh#BnPV{f72Pv{N_s5SG2c$_9sRplBl2Ss^|0Yh6q`O=bO8dHWQ3qd_ji|eNP6h#
zMMx+`@nU*8;)aIcKgGc3h*;-x(H-uTHHev@%uKZV#oA$E=i;DfIi)}w$tdaBK+FI7
z;_l@$&6C%qd$4M?K3ZWFRTq4#Ctsr2(`;XLYT<-q3A0fL$T*f?4P|GZzI&1S#)*~P
zdZ0Ogd`bJaX@bV{O#h&_cqi;#XJM`!o0ko*`LWF(2YmKaOd&Gw5tg_A!iwH2^-2AgQ;F#iSTI%A
zvG)PluL}wYGTh^L*f=a0xkv-DJr)XApPk5D&CO=J6{k(O*Evx+fdQ0Z+(gXaVp-SN!@qVg}6;v!8sM>Z?6rNIq)cG4mA#N$UbURRzDw@EW3$b5A4zE8b=S)-|R
zDlJA;wqa%HU8qaQW1UeOk@F^S>?TTW0}Cdkn+UhZp*|n_^V5pqCrg)~bxfbh@7Sfk
zbj)SAFiAi~5qXf#R@`?J;)>GI!>601Eoow}Ipo#xd=icng|n5Xs*9)U;nI(JyQFOI
zm5uTm4>l25(^<$>_x+Vzc}UhBD$V7wPU+mSI)9`?SeyR`B_CoV|z*pVf!e
z-5$?-^rbl60{ezP{3^EaJIk-PR6kJTlJuOJVKr-Zhb8^|@{8)Bh1Ec(M3jzZX6-!p
zL$o1I%A9oRz~@FJ{HC7{Z@prkx+0D&jdBsOwMk*>{U)WL)#$b-Pk%X}*`#IuWU9@f
zxFRz>)gVsAvnbyi(bZzG?ax(J_?C7DL<;KH%K{)>ivq-O%r$^82NMQ3(E!I2)G$C)
z=1y!q$`q?0;@rM{uWrY@u(1lP%aJ6{_OJ83@*6$rMetr}YgBs4BQD$7+
z;3da+C)u`xo5r8Y2GTMXJ((Wqe=#myYRk1!8pvdpKU)6ORj=8_)UJ^(td$o(`Y|-|
zWxWx<`wujBXAUm@{_z-oe+hB60m-l<18$GVM}h`FtNA`B;O+hpXcfDdMNCL`QSh|j
zfk2!ay?Ol^aHJwJP!QRp8?2wJDhnjcUrm+A1ve|G4p0gl1o0eRwpZM9nteoTa$P3wt_jm*4Gn*H*L?+v=Ll
zM)bQ~pP@S7;@j@-sU*eykl3DKNge!&@lNwXK-BY)Cw7o2`BOYa!rEdnE#ty7O>WXf
zEk%bU4ePwZI4Eh~2s8U`!5YmD?FD)fpFVZ>2)kh1D6U|z7+ui}2Lvw`8>@OuWH=JceOG7iqLKAlRc
z^e?IM-Jdno%rTDMmKoykGpE@eoe*&UB#at??U3qx5`!?5-+PAO3JCgB9kwTB
zC8wOYsESG*KZutd3&f7XpI)juzD69?r}vtAIxI-IV&NHQ&=?L>*Uq={GA0~G9uvc+
znc@R#9P&A10S}mIu<~xR`+A};f$X(85fhZslL|~^Be;4z$;nduOQ|F?`
z<0g;gr=~diW)=NjN4B+5%SZak^<-FGXXsLgGbU4$@4dv2KD@J(j}#-j8SoU9@cA7K
zlMQ?^ee<2)h|E;{TZ4Na@Wj7}uQ6bFN7-*ES&!kJ!^lmV>adIAv=(_0wC@2!_|2Uu5l7qbnC3Kp|3KJ^)If|8
zSJoFR-DItaYaHuc?aR6B?Y(WC2cp{g)HcZPcUV{quA{SLnoQDKTqlY2O!THuXD~J7
z_q;xDVBDPeG&SuN(}V$q1hlOy%M46}{ZQ{lx%{&@6#{j=?-6?hR$D=0G$mpL8o<07HPMtHk*3~@pT`|PlNd{P*^*@hmM}MUH)Nz{>Rsb-J
zB@4djhl&SfkEsJH-zyvuS}7Ozv&Zoa#T=S$cC|0bdG9?JD4TsjW39VS6N|Vl)xWn5
zEV<{y&`8V3RuF`NW43n(aO8S9MYW-Wx5ush7}bI)hTDeK=MvL*Oiy&Zw2xyho|X51
z?;JccQGc7<&zh$qyQoPU*%v`I!O1cg`!j9$=~`-{1-55%@&H@1s@j+KmvAyh4@ZZA
z!SGYIC8rdW2uP26p?!}dBv47=i2Adv7f+A1-zPM|3%wdfK|D+
z+rrbr0trDHBm_i|E)fuzw15&K-67qGq!QAIvzF{jKjzwM1HN`fKR#%5hxByG#WuD>|pxFSJ!!q_Ty)lMjHQ(w8S>D#n@-*Gb~w
z-~^au5)u+L4pUA&YJZUdfjuPlx!|Qx%yla7lG4%zvvJPLhzFv<3>}_0kf2w%`wl)0
zEh^X-*u6kSoBB1YwY7Y6R0Bgzb?CAWq#+*DGr9834B^tj+goMt;@GGziwX#fMjr^x
zf4J=ZW^IJ
zY%LA_^iz_b^rY>9B4pi;21etIw{LmZMnn}>xEtqW+F@Viz@|xFHeL71&$qiER==Uu
zu@o6U#%#fcHy!s8sK8lS4;g!Z^{@g|2pdLkoY;&VL04Tl?};Cz2>UDus_0QJ!(X9J
zr}BiYICrpj(#OB?>S^2q1{^ZW6}!^UewR6W^v<7;m(D1?{_@Gz`Dsbg#Jk1%v3F??
z3+#4{=ibdPNzHhVhX3dun71}d#PBbizk2?)`r-My5&85iYVQc?Pj?d1G9ISW-&_(9
z5a&U1LaOj#y@m)4AeI^Ge(*@uT*})iVk|E6M@A17W98>HWgQsb^`(0e%+&GdyZncj
zIFlwOdrtL^ZrLZ~j(Hl4`Vz#i%(HSF)(UXB2fW@Gi6@znA7NCJl_}h^x;~|7q3#hN
zhi1ld;czB6Dgp}?sGhKNCwAb#a+Eza!JeAH%+qx1ff9ymG4I<jd>~rOul{W4*u
zRw0}=Cdv8$W1>yhHR23hZ!p#QnU>bpo{&6e&$YZuolDHwr;f|dxAyvOSKfme$V-WX
zzF=~$a!U|J&#rI3#prrRdR;=5>i)Pn5TAX`1BN4~XthYmmdE^AAf
z
z4nhr}C-x~~?zsj1$sY&)eoNZm!K0G%0FAwDnw#
z$%cUrulu5ZNU;CcZ}dNvd3sCIQd6rfDG@@uA!x}2UVP2WGy^aUN)ZGd9;>MIkB!}>
z2GH%}qlPrCBtmLxAk;@iiu0UBq!PJTqi?6iMzi0BOuHI8(DB#!K79cpLT3^B7+!0b
z!sv63|NKp$F#0z>TfyEX9~tVrP0eywTMb|G_~IT`TKiFJm_GGJ{fyym3xli!AEHZc
z(L)|`AOum}2@174%#W;<3yq(U9}Y7Sp#KpDjeC=v?XRP)=e~cC6~sT^&;|FIu*!1u
z-a_BYIlu^@z~#J`@M#%BX{!|8|oblwmLSjCAV-n2~NQnga_?}
z;tPK*KR@7QHRN~AizB|_*Yu#ZVPR4cze4}?e|VE>gnll?k+B@qMLEv$%k9|+wWp%sGtqE&Uinng3
zd;ZnweB4bDzgh#MSQ?wiW@15?sdGtE`ttY}bHaF|3<3dL_^@)975arz%p4eMw{={0i=o3}e
z5}TO|LaIt`HkNX(stuSmB{piE>UW|&jNK!D_E{k+>h#{AucTze=-{NO;G<5%1gQkVoeI_qX!)!AQL$ErpUSQv6e)%2&>tbz9+Cu`SZnh9H5&_yizTST
zDk|g7Re1DBht=V%rxi>}YcdJs*S_?-=uVY1F*RLW
zUdHqhIk|uZNAW_Rrr$+hM$*hz
zHzqYbVx#=(X`4`NCE@|e1+1439w1$hv7piq5!4Il!4XOipnrEKs&HE>#Lb83aRo05
z^iqm`CH!fl|Mbmch{*joiV$}d6#>BqeNzQZ^sIYxcZj^cPYrp6=?=N0$<^|*ZU-c7
zH!C}4Rq1)wMF(``>1>I3V~-(Rqr`>13{`(v3BWC+#6s0vbveWB+w0|5V7=$`@+BKlK1&{{HRQ?>O%%Pt^JQ#u4{9oP|Z9YVi@vvuw~=qR;-#`b12GOI;y(
z;o<@W5Vvkw9)bPo@FI))#Bn(PfqD%dfS#;er%1yU|3u!k&@MH35uWz@n=QdSV?+W_
z8uHB2vdF>(SOP#K2o)^V`xyWTI#hr>V(97a=D7S@tQp3xtU`x1lt>x_z1W9P>p7tL
zK(C%`C-p^>Ld|k3cZT4vrJVBLArZ2tH249Ww8E-izEH<9aFbd(9hfa{Ko{za24w%A
z-`dH!E_Axz&SO6CAwms+>b`@`#8X!ZiF(@pKhN#To-SO2uAf-*VexB-u0EOpxg?M?
zK@Aqn0lP1u1s#JS3VYBwEY=dJXsiRTuoc98fz=FogwWw26m<-?2Q*xlzmJwj1pZHd
zUmu3occ`tCx`S>NA}yG|0LO`mi2;NquL)9m67~B~Yr#W=dC(&R14>0D+7A*c)u!Z2%I*JkEoc!snxhTm6WH3~1QocZvz51`{YkeZro
zKbqyKPS4Ip3m?F7<&~>{@+AGp16-`tszXcLCLE8AFbN7NDfD=ShZs?yzD)l
zQLr%*{h{$4pvy8dV_|rmI6Q_z0)V&G)J#S+1%c~TOJ5ex4Qg|R4ESPjXhDP^`K$2kI2U}@+7|N{_?KWx6lX?bgGSvTgRiv
zkJp2(Tt#n39!>XGpKL?<*r$v?t;8xQZ5F>sf<$s#j_j?6dQ=n?bh8GfL2)OrC*&Q(
zeqThVrR$an4!X$rAEP>aQSJA?e0!ea{@3J-VgJ`={?|{klnCFv5j@mJ|Flh!MR^YC
zrSlOARgP5DiXR9Ol90m}Sx})7y9e)}fmi<86SeK62ST1yRK$%7P<7Ujz*aNfUQ_tVtunE7oiUk|apOWg6A7wV%3DCa;0)3_<0=l-+PS(S4NzV2Tgi!85s>}oqS$9~sd
zH3@Y+jXP&K5ng(6N37~pcC=IQAmjbYs!WN~)`|pYCv{<8+GP?vm_Cb3OD-;zM;s&{
zfaeRmnX0O>B8yYV^CjKf1jg6U3vDnu#aN3x{QhKaJ9-K`_!c3N3=6BYv7}L|p!1Jx
zzUd>&WMuWei6FW2#K}aWJIHLfr*#auuf2SZ@lm?D#X*IX!#L2S7!|Wwcc*MU!f!r`
zW5)~oeASw9a;)gbIJrT2q1j)Caut%JeRN^gTLTsL@lDN?A0?_tFCu8sdCf$Hj+R}e
z^H1fxi>ia^Ot+LZ`rT3Kdo)YCf)+WYoF|V6qKTUCa`tOc?C;Cl*IdT!Oe1;Eqs1*`
zt66PlK0ZmoM<&31gX!AHcIk6G-KchUm8_7X{?HN8L$O)`+kw%%)~8jrC2z~h-?v)v
zlWIpMP*~-3CR1-OD6Z>o>i1`EZn9t^bSnG+KjmflT;ZyGk7j2)CJH--aip@`esf_y
zFV92@qHLP9^tr@RdUth{`4Gb~y~7ZT5T#jcyoK9h&dVwCSHep>?b1)4WUb#}!Bq3#
zJZKKnK2g~lJ{>v4)-FzT6DckXInBV}Sf`%4%K3TOT*0@ouVC3%=gdw1V$No+E#d7<&F*#Eb!tPAcXege=qf8^1*F9h
zYU?p+aZm1#o;w^NA0DB3?ckC*5c2LxRw@CMA;;Z$5^Qv-%*CX-+#=9r`i-R-LAJ8i
zOSZJGl6gISt!?kS!;0>wnDtc7-_I^hExk5Pm6mQTjOMqhqN6TL*6b%&B(T%F(BHqH
zh}EgTnSV!l=cdyHBYl6x^p&|tLNR3G=l(PjC8-5Z?zSbCg|@cKnk^moZ^OW^Q`$M^v~u0M)tp5mU4
z(AIcaC3EM*!veu>n5N$
z=<_D`W!>Gim+tZjygBk@PEM9iTF>ha_JZpIdPSp#hD?Qa-jq8zo!W8p24l6kSdkKD
zT^cz`ry6&u;657l2&A~UdU!E1F~(uFVl(P^poTihcJtNgA?53PUfwm89>;{8HQy-R
z2V<&l+H6Se6A!futW@j_&*zoC$aYvF7md0{v}+=}^JlgeC)As4Yw
z37Y=ggxSQ4E3+9~oi5LDZFX0<+pmzg&PatF*lTyNmi|mEwr`I8A>n~L-|U3f&h!2&
z<^HE~3z>-!p*JmA5j_})sRs}nAG~te)0uvGnB06Oq1AZ_OPN=ey3FK{KhdPn()t&7Hgz2K>C$Li?&N<^8{;O!%k1yp~aj(pTCai
z!h#j@_zPI^a2rJvo`Rr+~u2?uW#E49AM
z-d)DW9kn-o6V1VnAviUPm2y#xZp9)!HCoas`gnG!Y{$>?{-l&PqWxq*jQjri2gdPCLo=rJWdiM`f6F#s^M&t)JmWt=DtVn
zma2-R#p{#}&d3_?pWV40I-fJ})zxnMR-bHfUG%(ScR=vZ=J&z5L!=4#{RFsk`)RFL
zRis(*J<2y{&C06o_EmjgwO*yHum764aI0>F(XRDV*hN%-S{&^H=^p;tCN$1VmulXo
zsJ6acJqc;z_0n@;)~>zcksLov-!by=y!a3&AGjnpcHV_0;NjSC*JuI9#+ui$QOVb$
z#l9!Svu`T4-a9YcTRPZwZ56~VEBxua8sGW#tF}j@hEw5F*B2M9uT7{BAaq2APRDnH
z>U`HLdAd5Q7>mcH9ejvCh70X~Okp&^!Npw)DG$!rZ*2ekR$*x|bdWUR;a61;nGN!M
zzt68kmWD{Fs}EGSJ_`1}aShKRl`G*d5(^FXk;ziG5-QiW8||Ge>(sKBC_g^9)5k>_
zS~*+R7MHtI$Z^iemerQWWE6Qq=_}_M>;XDAXu{
z_Jx|Kd5nT=TE#`Qvo8#Sdt8jhwaxo`EwVE{bi9_n_)98m{Q?QH*4eoGjg=GGTx-Am
zXx=vg?TAMv4Mua6qeJZ0X`vta)VV`0o_PTEOwf_8Qu7oA76qw_7g>9wNl
zzzy5F-7(=)T($Rciq%f?`$<2&D(Ct2_nU-;>gicrOp
z<%L9XAM=&I4PSa3>t%;c>34LLmUq!t*KNA2)~vG9Us_x(O8(aPG<{{KlNNp~8KT$SsE8C7
z2G1lMT-b)gMmk7()8pvyr6&KJv3PZ34|T%v*;R8>lF~4tpbLQ*Zi3E_1qyZiF?Vt!
zn;Ih@unNxJk1gjZ!u8C19{Wy<-C#HMnbr%R2VhLs!3sF1FVcMr6%YJ?(l;#d9*=#qM<8{^-vfwD7_tta$CXGv;oYCk*yNMCJ
zdp53OdXrwpw?&IB+AAi`=C*M^$*}hmJJV8I9aG23PotKK#1Kp?zc(ebL|iZz^yphq
zteaIgYF%GkxzKNZC^=7Mb5nV}gU<(@61(1cskA<2JkR5Fc;m1%O|MxLnOWjqSz6?g
zs!!jf@Uh^D-7-t*%WJ!GW5o;!_=BAv^9#g~SJE!59#4KKXCxuiGT7c3s&-~L=y2XH
zR^&G!x~~6yQn>Gs?3JJ@rFOL)(zI_`*Ab8S`Roq2rkzPd@R@Jw=q(F&jM-hq0;AAv
z&9rOe4Wngz9Uk!&y9FCUZ?z{hpXn#Fb4XZn){QQ2H(vMOY$v&G!oVzZ-z0G6LfQ7i
z<=!7*t87ZTx&dAol{eKr@5}`Y
z8X`P<(8qNtUGMl&K}TcRTyf(ImKO2lFIpv<#SghA_>jBy!6PP*+%l;aCC5V|OKDhq
zK5uMDn4fbRw>->y4Y~auM_t!UUm*|Xu9$t7W%9je<@Le~IVl+5rAESZZ3CB!eZ^ds
zc-s(cvP;(!gN}|~A}6?%3P=U{RJqQ4L!x7b0@ItYJe|?k3YtENN_v)lkNM=Y9-mGP
z7uUSH;$vT;f;7NC>|JOJ$-BX8S~GSj{?6B;MAx4E(Yn3s;xqR*54P);OKYlGoP}#2
zW8RpTiH=Z6_Iq8y-YB-C&X!`Af?_?dKWMv2)6d5uz+->4rfgqvDkcE~kuCyBif8p(
z9^9{%m7=nDC^m1ivwy0TMQ+80)vL&K1^AA2CHwJp#rK81J{caZy0yc~(F{Ah*`Lut
ziDu^!Ot^*tYaXrl!%57r^!l>dsf;fT&O81Er#(u-
zA00N+f_%~k*9j4IuVFACJkm!h_VO$(h$T8c1f(t1VGwI}ytz;QO?|zRqfN8z4ojUP
z23G;K_3m8wsl9kgYIGgy*V2-&6|vQa)aa1RuJt&bu(ATtq(CGi*CfWpvUR0NVY#;F
zO=6UG+iG?Tw#J8ts49W7nIIxx#&8WG)3R
zkCMI|Qj6QvUBzD#*)o5O7APX^%`yc?Kc`}q-h40E`$1j3PJ8J~LGAWJ-FhM))_k|@
zPIJ3NXmF`AdTH(Rxg3#~B&-#=gNa#cJimUB&+6t!+)0X1ssQ)N*+BZd*zF`)lv}
z-S33ZP~%~$CYG6zHB*uVM%9IUbu6|Z&WJ5vIj&WY(qQe=qBD6VMzTp*8ye5NdTm^V
z^Ary^GCWh-IH>X|;0S2+%X2j-kR_O6-Z#8$wivb={6kRxA5dwyuJD-AQG<^~&~Mav
z$zjymcCEPhy-iUNjsd2$54)a3&eAouu`y`Iesa?KMu8N5e*3BP0gn|6eV2kMUjAYU
z>{bH8^dCt!_#c{#f9|SUb5{yJXV90Jyd#^DzwxUmk0|$-$2HiGx5f~A%$pDKwiE@T
z#xzkM&vokSw|<(cAOd(ILR^uatR5Fv;N_|=oY~9I`Vhh$k!lKjci+H`j2CDi939n!
zObC6JgS@>Nkcr1~+fd7<@VFw=Zdec|+%NqL<%Mi0#wEK3UJe(%+?iJfL1Cx6
zn+D3PZ}rXZrV=NkPb1(g@Yy@T?Xw7gt4ea>7U9%-?X{t5wTHyC(5dLxC`sCl0!r9f
zk_JYRa5UYbS1B%<_d1N1wVDqZD7rRVmU$5q$-|;PXYR~%FZDt3`^z44+AT6UtWkGW
zocFs%XHdga75i%T;G$@@^Yk^v=SjTArhu{6-1c
zOSkEJT$KR+&@Zh-Yx^Y*#;8~@X+p{t1XG8TrK-ny*D0^se7nBp^Z*-fIG8Uo!{i;x
zq@R9i)a!6WlYcVY{%r9`PJRLX`hmg+&|oPj-~d>+`oqWM&HllTpe%_v
zj|~x|{u{4&s9RTYp^x5U-2JFOsnBEvHGkIaV&Kslf8NLC
zF!aqDPi3A{Z_T>oy0Kd`epojUpx9KlgK@a7EkRi-VKzePsvhz7k?)p${gAiFOQAiQ
z!`S-v&al@t56=*he#+us4R&26?G)n~&0h5>uA+bW!Gy<|*ZCp^Fuwk=u<4ulDH~_eK
zG7;)K%-(V3Y5y2b$&@9rxnw~T%O%%e?bl_F<9|Lay1(~YJL=nW#~&Olr9U8$zdf?Z
z?n7_FqApZ#5hyD!#y;!0wf$atN_3LP0ndF
zRIO|3g+Eh*FWGX3VpIgCn`2dHhnnl1j3VQV-6m6im1$L&yBa6FieSHSm*Rx~v?tCw
z43>KMGe&S8j=kQq@WDWUSIqgp53v8un`d8_ycHP#f@OU~sQMk?I5Rg;hZUs4Zwq<=
zlIQ0!chU1Ra1WAP_VIY&!Z)+Q*5Ys!5c1e@&4v_>ONdK~^uK?l=j(qxx&QlknP#lp^qQ>N1rtmtyuFem*tFW`n6N~gVWJhLc|)u
zKc4kJfrtKn*8kEQG`68bltKZ&Qj#Bp0{}b&cf@Km_eA958^0L9Hd&x58$SMV(F!35
zge9s2b>-WUiz4t7p57p@0jMSe!&>=OTreJe@2!RKWcq_@{F4mxuk@Jz(kFG&4(Jwn
zpk}A3VrD_)tI*Vc7inPof=#wN?JnX$7o&+ElA0Rt7yx3(oBg3o9vI~yD{N$JeApuf
zK;Hi$H|1a1D*vTV?(|-ueHwKOAke>ejsa@4V+zfWF=7sF-Z`u|pk!1k*o|aZ$eHRb{80hdyw(*ir
zB2-XUZ{~4DxWYF555~yb|1Y@DKe3*a&fhwt5AN9&SLUj>`T4=I0v={+85tbVjtcOL
zA(h)WW|iRo5q^Kz3*tYWjl*5`JpCX_e&)PHGIDSawz$aJgMn?>KHRhb>IQ<9A}oP|
zhb%{_b3Cm!_z*@QITGw-)-Sj
zv;t7yba>MeohMUAmZyzh`92d|_dA)dj}Z2T4X(A8OgZ3vD`q*RbZDNU?`8wN5LK`8
zxrD5iR>Az4nc@#&E5U7N@FKCzdBObnj
z3Qv=4teP)wOK9STni($J5gFkP;5nSkJ5CiyhbO^$y7eW
zy#pHngy%mPE&mGR{j)p&@tuEu0*KwskLl@+!1(~LMp*}DxYUD(;O#?;2N6T5%ApfP
zN`{B;P=iKD0aLYLQX-!n9oe|octN-uG16)X>oGFsfJ^}&f%f9fnc^qiq5nU2z~wP
zjaOHE#RmrjN<1Q%BlVl7LI=9jLi3NmIovvPQy}GW2hJ6rhuuo1F=q_n``-+}_aJBK
zT`I9h-RRfO+%`uGr=xog%ObA)0*G4pHY-SJDMh0%jj04a+X8+mMIo01wJ#y|7=wSu
zk%{*@qer(RWBHou+KP5ZH7SF3S|}^!BHM@z^`RjP7J?shV|YsaA&LeC=NZH{wUSya
zRUCCR%y3;Pe7Optx{39%icL^td^GCI>R5pRnAG_8j!&o~{;
zio(YxV>J~w1KUNQPyC4?Cklo$FUtk#iAC=&eK3lE4099jlAMt;V&eMm+x+GvzWrk&
z@q}gS6#=^r&#q3#(OUg0k0XLe?KeqXh;RB#7-$N`LkzBL-nCd4IJ^}s(Jn3wc$V?3
zJJ`7w?@~$`Ct?lncjRJqR+7t6%VUIj?iuJAm_e{@
zZ*PNRGsrV>x-CsYeLq-JQG-<=F6KeuE)ocz2EcUvi_IFtd2kFPpY+=6UQfg4FtR-(
zfQY$*LOGJvSAU=)yCy+{$S$;suf1Day?O2d{{NH3|4m}zDz98dCi>#mRh8a9XifhM
zZUMRzQaCo>EwzN)K_U%}Qp;{spUOZ%#~&^LE^e}|)%=EhYVyj@Cw=ed*~@iDW?{F}-q0
z&+IY4ssWqwzov5pP!ydAA%Xipq6PjdV=zxa6CU;UENc9^^Tfr@sgBd*sHaC$mjHEk
zca1uWm^iwDawsx)#76aRJCtQA{?Lt#PYy^;-@n=tnW=2k*kIf-Bv8`s=24)ooe2v*
zyOMhqTfq-Bl6)fa*p$7ERS#P;9tCs_I4a8%E-sc~{b0KptV&RK;ZBf`OfAtSk^Fz)
zV@%GRL9j+b4D0Caohi#vU<5Tg^klTwvK|{XuzD_CR{l1z$08f{T_I$STJGO9@`ltr
z0u79=Ibe*A83`SIaM97Orgtfeha;t^R`G3-r6PW1%Q052hewz+9YfAjhKG@apVH{p
zQ^-3tHRWn9&mBAUwgi03F3u$8sJCvDtR39{z9;jeG_|~Z*yepp%R@AkGzoH}A`9Hc
zrY82f1lM2%nESN+{OZoQzs`L7&}9fn2jwx
z^zvw&aohqQ&a_5d?*R_;UFi7T`LoAz9}ABs&aGpLL`KB|UTV~KqtkMfZ7x{Fe(E-E
zC$sVLMN0HOA0ze*+d4;3t^3Af9>J||tSwy8#G4ttL9xGS0d`}9DWs%GT_zPPKN6sfs(@p}~c4X4FQB-aWqe_^@kSaV6mdGEt!n|fX#BU8TR={fx!
z@FrfkrNGw?+ys}J@U~PQB+|+89*cn0sQr?)@iNMmyL!dk)gfjbQl*>0Xl9dV<1f_c
zr#N{ydKHs-tLV$CtU$8YBCyWyVcad=d9U>4@y2Fe)Zk!W8j6H>j0&LE(<}qSpCr{A
zKL#~nh}GC>y{KB%i*Oi@Uyq=o!uirSuE(T-XLgE86`nKU&8;TepYiaiK5nl(U2`}b
zq_3wzjP@Io4XvkC9vcFWfSOBO~gxXJS+TIEsh`)rQ!Q&aom9+Z28Ju;b)DnuP0
zRtF?@m(6s~{`3lB`jYs(7EpzGo3%{s(6ML3e^DmF0bJ|Lb_!as9ZBZuJQo?4)?|d)
z-4!FiQ)TepQji>*WJI7G&*H8cN6&DpvRM+_OTUss$2cjr@3rl$_uZzgT`&2vdHiB!
zR_^3fev%W~AK+K+@&=ue1zkq3{4`$@oks@u7Fn@?x6|Np+yDSYqa99=n2f%#*HRhR
z#f!B#)JgIQl5%(9NC;drk^=E1x7?ri6c5tOlOE+7MTBKx2y(3A)c4uagDo*5Ko^8Ab5dn~5R&`{R#I*UUWbbXg5a
zPKRli@fU~Yqo)E7Oa)IRQ!8`VInq;&E=Rgy&c&Y$rQ16`lx2*)V%t*{1-JPOS?^Xf
zRG0+6)1uBvmhS=8&KZgk766=%2Xd99(%Fg``|@9WB=RlE@hiEwn@G-CyZSYG>*G1&
z-5ma*sm-fxf;A=S#2m@zh=YDMi<=CML_Akn-$LrTDW|Iv&}?y;6^)^Bi0<_1IeK?r
z+%3B7ev=Enmh{xrgM4`zLu#MW)Y;h&E8m5&eW`pF@)3RCXbQMz#Db++=J9Z~Msysr
zGaog(Os)HDrvMwp1Bjxqm>7oCnVDB
zu=DjR?fpA!m~amLP~S4rW$y#}3E~|r{i`D8GqaU>xah`PWKOrf*L>51K
zT9&*`BbY#q7ZOVNh|jquQFx|~JE*5M?Wcpj(0l(xU8E}Cy6ETt>k%fdTfU%;YnfCl
z8Py>yZ#6d|dw+g&BQUXv#r486I~eu4%{Mh>s}+@YzVs#g-Fc
zQP(l=^RMbE_WkAhH4U}C-rKoJV=60B3oU|50@lx*B%O~`RnrfxIo(K*63MX9?ph4f77$nVFdxID@60G-K(!ii;cWAdc9(4I3wI
zkIb`(`_YfN^&JO;1_`eTP`l9@R~K_Wb9~UWFqFKm&t;5S3kK{U@59Fe#l&Y%+~T;1
zu0|t`l`)^DeitBXrcs)4)z@Nr3Nr-Y#G|M&feN9qv8N>lXQSCMfMBp=C%?eax^A4c
z)Q?Tps!MPAQ+Yni2k%hE@rC^$}te0-^C=Tk?F6Jbkzdu*~7s-B2
zkNRUU;%_WsCL6S5UvXdX$e4Wlk>qpIvzP5|w-3@-B`ttWyNs^O_=YnXc-F|EL2@cM
zJlxNH{M$t@@fp|MZ(6uE*eD*BDkYfTZlVN?`r&jz&R2mhD$~V);g-r*CE}ec46&gY
z$bpSDJVBqDv(2m?eKHiSqF>-YMUruxAC2s}c~S>~sRxdvzvqt4nGY_zfB7lk%23qg
zrI3*|qLJ4GT*k}iS2QVU$0klEAPGt3MsIw5ajMf+7I5pse)Y@A9H?n+YdbVN{Yp#1
zIq#LnMQE>gtHC+1xHkIPyvc8>6>1|UeE
zb5AMD=IRrZT0pjA1Lr#w7H`k*w{ojv?ic7hY7`dT`0<81DIhKLS$A?h>bqOn%tL!Q
z@1{oO?(4;cbrn^HB-bb-QZA6R{H$8$`BkxzRPH8|^&^%5+T2*nwQ+u&B3GloA8VX_
z?B>3n0H2pJS-pTGZQsxL(%SH_6Rv2Y0zB-`_j@)QPolkSr?A-BC-tvS28rc;`6f`J
z-CxU7^u9k=cKpSggN`$9Wo$R#;5zTy;RoSF%Y1Ej!M@dzj0wUMrJ32_17tQzGl(tc;9gtDboBw$y{!+w*(nixEU+DT=+5^vLSRN>lO#h&$?|O?N$7Q2it|nhnpYJmgUxd+f-@|
zWxm_dJ+(E;hmA>+U!bh&yfxr_nEf=4YA3EEo*Y8ja&?strbM&N`}kq~-kcIif*Zjr
zE1jmz#y>Aw^2TWq4pXFrk^ASK`PkHS(Oe%k6Ah9=uWRLV%bN!e#y${EqQ<8HlojC4
zJz5*2>O3`Gt(tmQHYVhmNW09eejZ?v-*?^mDRHg+j5nM
zGaQOh_wp=17xF$57W>J3=@moMTPhdY{bQTq!*T(h%RE##9qos?(<656&Xye&C^uVl
z0qf_hzfZ3iZe3nx5}<5dwo`SR{_t7qdYk-4s}J^c_X9O%3$=1t%Swle
zA9c^&+}WbT@I5`+523*4FaMHG=6m$)#r9Zy%Tt1|*wU0ubuM?3cZ#qv4_dZiq(YL|
ziEC-35aKbymBET}Q+0dXBcC~NRN!Zo#XeSFmCrkB?vASqn&i$wI3iGE)Svanet~Mo
z^j3cUTf!R+mg5mx{>SubG&hzid8DtZtHI49@L+3*$dJ0{5H@EZ#gahpjZ>;Y;-LzE
zgayJRts1Ot0;8ISa9vJ~WhH!;NCHOh`#pVVknb_ZNZj^FW_c)}@D|=e@WE%()+rCuPR248k3CaJ=$W$ooley84@9$y9m
zDPV0!Al~htBE2;3h6V>$f|9?2LKnyY8?wj35p`&&JotEk4W}|^80fOkC3=j3{~ilV
zJET3GoopbF%gNaTrv-zzK^iqEAfU|s_yq{VN=f-cDI;ixmX{xb94FK1sB#>LB}DOK&!BOcv$k
z#d4WF$$WCfoeU)E7eVV;<}{wX39RBb##~mhNRaZKwBD{@jk>-b5N)G(`{+jZO_3^4
zD*@|FD8t7Qd4taF1sY{!40v)HDMlja3>pIi{hl(h7vON&-8ihEdNv#~Us>mMe#Ta{
z>)gV2=0=dcvn~}bZrG>DflsBpHyhM6OY-Eum
z-YH_8)E%j)M~a?Cn}9Lg95TyMYdt#v+mi<+^Vi+HPb9>
z@P_&=#L6khY2q4q4Ux+org1m-^XC{qh(LJuG8LYG?fCdO$j8uygPbzjP_o5I5d=iR
z5XI8myhVVInpzKvAJGmpJ*l#!PENZ3;etZ!?CdO_NK0$0N{+_D{9fXPB(SCe1#3l3
zP1#Fe8vzx5g_V_Vp!^D@qi|UCqi26UITS3IB`+Kd8YhBb6F5>f3=XOe{zQ9yfI8?1
z9v1aNKMfyF@!MN?bb36xl7
zWM+c>Nfr3~p^d_3;NTYAbo3uS#H6BSWwl0)l#ZN$mU%1~DP6jZjZFz66l2xZrwub3
za3qIz)%3;zCq9rE2cwBRW2sMFAUP}jsf%aa)edmFv$?Vbm-Yu%Z7_bvAjCZ-q1_yF
z7EF}th}P!<@2wv?mS*W@pz{xk$V1TEv7|;(82!{xzGYh)a0M!y0snniYUGd6iXQh4
zH|e}V=Ju^KhwhK_C>}1V6cBFJcH53bQd6NFp8{mF@Icxu}
z;O1%ARR+EKEA=Hv@yD{Xf@n23DZGC1{&6aG7y}5--@NIGqoAdg0M-pyG#ng?jFEUE
zpfBsK9fyItF$(_8FsZz83UAn>yU_EV!oNPmpjJg?C5Y(G>DKsycZFY^VaJ|Y3>%@6
z4@mXqeau(I2!7^>9K8l9x4)R7@AEUww%uye&X4By
z1Y_S=LNGzlxeC6B?|7`hgu#w+x8?0|`A5PyDkGCM=2TU04ck&zFna_^b08W3*pniH
zUJ5)9RDQQ%3(L+GxH;W+ys@iPwcuLr!VJa$C*~nh%%_8QqRE+V9;s<)YQ_@Q)zv9o
zNR382vRIWWMsj(ATPi~`_{VO6jyj&mS+E#V)>Fv##sP)u*$d^?H2fGZyTDQpJQBco
zsECd#1^i_|qh6&?#=&75bhj^2QHM43?+TH=U>nziw{2}G5bFm2)7FLt24?2MqN1Jc
zZLG_eou`6%KKF43zA*!7e0WBDe7
zLO`I6242O-Ybq&4eEMVv8Wo^UZwcn;mDy<3W;f7eRH?hzoo$F&igeF66g&=?@z@y~42*^f*H(-niw@13(AQ5=V0)7ih>(DU(5c@d5
zEC#DCP;UoM?=`3-7Xafo9B}B>w)Bm`c^eice5td$JMO+S`=N!*rABZrYiT(G&3X{n
zz7MJbD7}ippf#8l&`7tuF)0PfWQ5S&#EF9*LRpE12oAU{(iipBFrQs5R%N2d}XUaxjUL!y^fY2Y;6p5
zbiPFt1lBV&-^3ANVqyZmPi#|DMlkpomtPAmE#3d3TeIh)&aNE|0t+x|!0-_KI*Y%2
zfgX~RlG44VS!SR9<%>1=c`ToQJ%o&jjxGl4nNkhZ+#HYVp0!~bno9A&D{0h4#n9Ns
z1)`^}FEch%^9*|WgLmx{
zEiHE#YY-kWArwd9LNugw-}?xLp3r14cEI!Rtk-W*bn>o)%FG`HpS7TKW09hOJOx(*
z`|gX^B~TLBghl0cc{afJ`t1t4Y6Dl0`)<50c_iJ-H^>oM*8{F&!S|v6g8cn;Qz6lD
zW`6nw+^FAZEB>@A30Tn0x_q4uit-^M2LuG5-Sp_`LH7F|H@73iW%ItGq$E8s1og%N
z+ago{i(!)ZBqY+8REsUisir}+n+WXU2;P?Af-~0=%o|Vu2L%L@WqLH}21b3(s%C5f
z{?0}Xw$NI;LZ6J7n2(p&IgW@A;Quj_*!OvO2#h^aM}WfXWu|)@-ONw)3o9V7sF22+
zfPBk12wI|DQkkVw
z1q^Ml?=Ws+W?}-%gX~EXXeK*=XM%!)z%m5w5hQc9l{zRiRLFiFhsvPV4>u$26wQP1
zZ0lX9yyF5rGX~1<1Uo+jGV8Qp@D1=?J>s8+Zv#sVxC756*13;g-p^01B2z{?!Yp*B0gFxU
zJ@0oq5>r|{e^^~&&X@yBsiimu12-x%@)*R}ZA-x{=ycR+oM3<--JtVG&GGVlmLkhh
z(4E`zm$p3e0cqzo2zhU07aTeh{g{ZS~`>oGRhoKW-DIGREw1YiH*=m>=cZh)Td%o4rs#
zbxRwPcN-fvDiHbei4QA;h6k)17Cf+bEl@vvG{C~YbI8EYA9vfJI%99(lgYq{w-GKn
zAU&u-uN$zI>uM1Z5w*negpOZzcoJMWQsxi{@0h_rR59MKgu<@%^^Mc
zAMCw(IF;@DKDtsQb224DQ8JZzsANiHp2s3lhDfF~m_mkl6l7azb
zMu7&57mW4w8F_|af(DxuZo+p@FneuBd8XhTAv`MKD*B_yP{2rLI#CX56f^AKRzDdx
z$I38oa609ZzJ&+mY1aOB%t0Ac2|V+1_+I!VV#&G|0E>e`Befa$Esz0@tq2a5JbkTe
zKo@p~N}=$PEw?Eia+9?n+(!z&6HeYYLPIHoKLXF*>58oB=!pH|dF@2?TxGyUs?Vj~
zQ@dt_;l=ADbvRiu2xc5lqGx7j?XqTneM#6iaFR%jNv(kZw02icscxn`gbVJB*n3}#
zZ=vJu*Y>R9JoYQ6W_I^dGGG6aqQ1@Ub*tKQ3lEaFR_Oi{rB&n0J#c^>$d+PM!ZiK2
zLfX#>1D47!QM@0U0n}6m)3!68w@it_npE6?m=oe6aT8iML~fy^Mz^K
zw&yrMAuIW1hJ?sg>eJRRmDaye>?%b1PNFqi%el&6F6;P3r{FbE{49Vqa)G>@9NRJR
ze_9Mmm`DPn0Qjx4{VH-j@6_eVVrX^a?a8=|IQ4B=?~L=MIkJn37p5yZPpS#8!_aQ5
zY&a88@?f^tQh?FD{pAa9>*~UMuD}%m50T(6G%wHpnt=HmoLJym3Oib*(A3oL;mm?@
zW)wLH9jk!OAI8pnjE9M5cUKo!alXBKkN7Ud?W3orPB?IiY(59(AJAEd{afzt?*0qG
zsf9%ao(Q~aBqgIW#c|fH5Fth2IhNI)?`5#lV*j~eZIhg^_EO&KdG&YwMC#+m*S7m}
zv)sT|kGOf$R?G|n8}Si^T~R`&&j{zxrIT!lSEs{Fh0a`o1t(?+opy1Gh=_pGPn=Sw
zVSS!1ye|w+o;<0quS9v_;>8X)pj{BABJ3M!wrn}&T+v_E%g+fynK8g
zAIRRl^ct51Tk5&n)kHX`Z#jK&%K*+KPCWlGEc&W5P8;~1_jDwmgCkP
zy1YRLp3pO$yfJzbA-l@v!o#l=PHXw)p~GIAV$a~f`GU_!iid4A^z-Kqc6K>AIWO41
z2nlI+S-;r2TNTb_-(foqXTeKz<7O@{dB7e|o}7kD8(|B6?n^iDJQTtYWiIUE;o*Vy
zLRz|`wl)LFHXJ6#dR{&a-*V=EmaPGnZF@m1)@8vy66~ON@80dQAI7gAoF}k9oSy%F
zGI~<&b?<+dD796*lh6qfucW!E(lu;K{L4vBwKWwS{;z*KHyHW#mw6!I^ZLu4c0Du0
zza>fjKlQ`3@rDBM3)>Yn9Bbw4;Qw%^tXoj#;=EyU=)60rl}R
zeuS%EN5I+zD?C&nri!n<3Ms-UNQ#(5R6`>=+?hW(#>!rL!=sXJuje(cf=vYYVZl
zl<$&vK&8<&|Vu0maA_!jJy&VRB_RxUCrHS*jbw)Y;zMqxH>1N?aU~
z9yV=!K^%v?bv@Hz02y=}xm`ZrIYu4?2VtRo`!0>;zN7@Y@rIjlFAE%ckZ~-W3Vr_#
zHZ6WiA`iGWS`;J9AorZvAuS<+Yt8+bp6;@YJOpO+zILZi58}&;C(gl2X7TmfBdQc6
z8SG8c>jrd!FDE4r=D%Y6=|pShT{T{0EC1j`2Pl2!L)-aAM-2OVjW;+O2x?_VHA9tKDl~Q-ZAYlLsl|
z@npy{PG*>*Dup8bv$W)R>5D&R0!yN!jSzVaU=F5?-K?YrAssTGu6iUngctMb{#IWV
zWgP8@q;YpJz)QEIIoo8v67>48jq8uuyJt`7t5<7TV4}QFmJ7`}o;ByxaU*fU$(5tN
zwzhG+#o*-%oR8vm;uFBrR2k#p_ikYQZ$R2))?0Sk#)a|jq8r;;55PqW?peQ*lai9)
zCdH^kMMdSix_mLs_1$$^T+^D@kDQFmvuODC05YFFefmjs2AD|P2fw9-3re**N0ue9
z5PvlhjzzN_R%<5r*dTFNY8x7yaMOcsW4=D@a2z=NO@-iZD4gJ7eU&ifyfHLn`y+U5
zWdp>1F6(^ZoRE;P!@`W0QTRk@p>hZvyA}8PU8Pc!P#BznedsnNOwa91x6ear5_hTd
z)~#D-j<4aZ5f~wY%*(6?I^ptLx-g-9=Xfes81i2jiErr&^RSr-dVP8dcXZpAySYWxDgkU7Q{3*un$PPQ5sT1=xVLy%t%!
z9(&8%_WspB!IK5uMvTBDyQxS>`oS?VDiOZN_YNHS@o#Z)-?@){eHC*&TKC25fLA_@
z!50jk=DN&zCtn|@gvZf+Am_I%-7k=e9mFcG#pgm9_OH$7xq
z%aCTZ@{1lxV$f|EctsX;Fq@m9|R`Wnt-lymg0)Fs0kB
zDhg2UWu&$zZhG47%vq}AxN8*^6$uGFu%9j3L6f+)0N&eNKs|(#2Trjc=C<7feV}+S
zcKy_&WhTBL-0i}<^e@1btF>huY*snkd6hzggWqT9J&Ah@Xe!pS9oAl-g=p$-*V$V6
zF4|!pnb*7LMMWpdyhRs(XWfRqmt%L~J|0_~V=!~*{R^HB=`N>k4j8O$
z23Ro1vEmYVBnNEj?(hL7!og2{BU0!@fZUx(er~1
z-5pMN)WOKB1j#p|(EUlU1sG)~KGA)isNO*wdz!8nnd?5JV;r_ra?)Kfg?yy1xsfs;Zga;?sf({kDk^-_WILMJW
zZRX`o5wbk(y9(^6=Jz+Yv=9`%QAt6;^y~SrgC=xTRP`S}+WF>Jzzb|t`@SP`J{t|#
zJAP=z?R#<=N2A?b2@=PL$hoL4j(482wXLF8=4C{e%UQAeIn~rZYm#zqN8P?%S?CV1
z7;)sV=UD8$dqLN)PYVJyAA>@fOa1|a7*>n#{L+t}G1
zF^kmNx)C)7(bLscH8lwSGD1S013K{(6p;z?YEG0uc%2>W`C}#vc)&U
z{Qk*F#2|P2)RdGQz#iF@g#PuPJ1Got#nbX{-)yj`KnQ{U1taIxUr8(xSl!(U3%f6U
z{-mB>6qrV!JvbhFaPC`sufmfMm5pUAh#FV^3{Sy>Z-obV20w#v|8nICkys%(&LRju
zy@47HPBu>Yt1^<3vj{*~t04VxyYHjfD?2+o@5wq4^*;K4!rZ&tJ=cB*{+Jy7zkt~F
z|Io<3R%|BzMZt-vy{*4;75`VnyZ*QT{|}Sa@c;j3?#AI0Yosx0vh;49Ery1c-XbGy
zLL$y|V=Cdbz<(Lfu&JYCR%~o6OeXR+Zt!QvAXs^yzlF##OX*S+mDd-N!aS-9ckbTx
zK)I7`MSCSI47ef9k}f#MsogQ~9LT3a!U69(cJKi>j#aOv}TUx;SkxnEzdpau`2e{X1|7N
z1Re?q`7t>LPPOC5&Ee^a3BE{#9CLcXjGr%GOhYa+%fObJU)9yPgtwxAd#2A(y2*8J
zybDE7&cl}tXSw*exvf#49T*reA8DSQrmH-rvmHEKK_@FeoWwIYnm)2rn`#S
zg-1qy2KfKuPLpdd2Q*esSLqy#)vi
z@I0lIH=9kror}X9d^%3Kl-V;Vt#9185!JTbb~7B#sApU6y`b<=X?=Z;GDj1i6{xbS
z$>>1RHl#(TPMwMsIni7k44(hgyXzBO&yU^Eo05!?*^vWt-PaizYOkCDKFW~}BCl)i
z?*-mQ!38v*qcm^-yS0(%Bwasn-uUDazBuVA_P?loXOJsyM{h1}AigWCMN##G=d~{m
zWKGQvO;3(3Qk$aCjl54zR+jJxm5h9T>M~4LVPRg>a|SgRZ|?>00-0`*62R~jxbhW9
zWDAQUFvLdH1f7EFYHCT2L}3(g1NfxF#aiL++S$+=i#Hm)e1c=T(NniH_Q8YkU%&Fk
z_($vTaSJ+pjA6DwcfdhZI#I_{ugr~K@bdbRI6OXXsjojRSJQRiHMl_3tyg7XOxl*#
zrT0X2s%IPtm780^+rLOJp)mNK=(e(SDdMz!ci{+@m4y7VZ!3FpgYk$R(f|E&S^{Wx*otzH3{ckEUE26E6dcx4;v(|H5Ywf^^f*u%K!px_%l)vpJjbvGzLa_bHhX$_
zfbc{@R}Y?L)SkKyRD#N}W9}MPeSFAW4ozXwHr;lyAc0;avoCBm%C8^fNsmD6=vKJf
zd{u!QJh2Z&tjnFDQ-HaKbzllJH@me#^b|t7LzQ`Ds)nm5V!VP~<3VY3;+o4f=RgB!~!~kS5drk
z3blkKA{y=tpf!{Mcmb6?$F6$$a{kYs@wRLeP!ab&(w`SZ2U{xq{04*pkLli$gleaRdOoNnB?SSPt}_@rx4*0S=d!fHKwMfB(EQ9I@b{f<$}S)
z5~zq#DMnV-i%S*Y&Lk-*UTITKg{PMjtDtZUAKd$LoPHWwl^v3ft*E617dE1RfQ{z2
zJljz}N{|FrRux%U;li0%BA-5axw`5~_v0GBl=~!DOuyO81qwSj3uV%B46VmoSy)-A
z&mR&My?|T*H2>;gZ!De+pl`U@dnvf_2wvxRY&Z=i693GiuPM;90moA
zdLG7|n^~9|?7mo~vrJTcMy0MXpBmk^4i=Q^zPpWfNy41DwDj^wb1KdaA4L4gwDQFt
z2}?^$M3OzA7pBtu_~;FIX1h_hw6wHoNq{LP%8V8^E-vMO4O!+$Bv6%JoMGRDXMnAB
zD>CwBe!iZb)}Fh1^x6VHv_R3yU;d+JZD+TTN!1pp#RjPG09^0o%cbT%b>F9vuAeOR
zK!R*gKPP|A9>OO{q_n?IyPjzH@FC^6M1m)}pn0N?m++yqc%x;r$Bv%>2l&3n9-2PE
zg<@X6(4IYK5Eo{pIFrntpIOpUC;?gPlxUxUXB!^=Ev68U57?*Z9Lp%vy3Y!Oz`lK%
zNlE?|j0_A1z-tmZg!~0VOi4@2YP+Vn2l|p}G*Ufri3wodvE!h9
zV-f^xAGKk`79+gl@MQ_<1{l(n{~gVin~)F`ydeNr`KF|YxY_!FS|AY+$bUpF$I=Zh
zE&fa-JitAl)8t^~Be5*jTcT!y8ABMm2f?}k>XDsVD`EKSww7%PQ*j&b%*TRvr?x1x
z7j+v_N|Is555wp0YpFe-jjDb~>k(D)j|Ci1&
zcazyayWMSCMkqOfg2mY%>sQ;^Sqy}O;EMVg9UUDYIU*^s6@e2L@6u@}%D=X)y8}xB
z?Ogk@ZFhAD@x{x=^&HnRH!nQ^`(YI-GwgM0G;ka8J
zp+wNIKePw*+UrO$y!XHX>%Pn0-b*Y0ep8vK3hm!-^~^fp?S+59eny=ga97uCTl#im
ze{1*)Xgmz5VaNV>RCk*o8$u~IqJH~1$ecjmG10AlEwH;0?wz5dgJ1*SJG1Wqfo_5a
zWo0uxC%`fR));z&cM%vRj^8Cg6yp9OvORzPocR%wk{N^{{zHP8`+~dTELw|*GCAog
zU~OU1gBWmm_=Ls6-HEdw<5S0ym~;g{0dw|rck9ohPKX@^Efn)347fPp{zEgk;TmvS
z`X><$TC8>{<~WKQir~2C4R;e1ippJq~)%iS?DfDUCkm`k@^(+L*j?_4=cx
z3cpnxJ#FdddtOBNJ%-lfZ9Uk?Bu7A4L+=c@Nu+B-ZFRtCFNDgyy$`aa0GD`rEC_{S
zqLV4^UbsPq&VFugg`;+(VNQ)nib{F$fTThS%c+8CrD!emMW7)adjvDsd)Y
z3PZu}zv>bQA=4L4Jl)+jG&OU^@{0Jy#O|yH)YR0B$%(|5UYHKKi@8tWOmeDkt*rbz
z51Mn-68GoC3I5jq(E@<#%{_U0PvISVHc^|PQB|yfxi%B4BFDD+s;VH1TxgLnefty6
zSpx%uA}@6{3Ci`Npg%Y8cD)o6i1Vz<*L#V5xH1Cj3HEe5q;AS#vrW_sAO|%_{0i<6XJBLIf4$v+D201OhMS|kucPm6r1JrtOvTA*i
zd1_UmZNzfpy`G*P*{fGq7sdSc-atjzX~anf?@l|3P%ToLRU-dsqxj=dF`el_%ujlH@u?~-pid^q~~b9Z}tej{%P_I$N)3PX|O
zqHb&)bNl^}yPF$}h-IxX)VPDc_h8yZBG^gc5kxyYCdYG$JLuVu%EaMX=UIWLR;6Sj;mA!lmThq1g
zg+ohqXQ!!|-Dr!rxVTeWR&+tVVSFWaAy9K}X9f#I0tyQLZ{r0`@=}_mk1YS$7X2+V
z%2epbQ=g>q&*qZJ_ASHh{Q2KN<72$M!4fy1
zkXlXI>9bcWKE+j>of%~lzmnUya|XGp?0ZJ7p5DLT1!%K-*DeVsXQ&G$f4!$J3Yicl
z5bvgez3g9MX7uvwWsNUhKnNL(%wg9nR{q|wxbx@{qOk^Z#9m(BC3wE{-U0DI!*;;m
zs^!xs7f9!>?J`cu$Oye0uYdSez~$-R$hr$oaYF>&`uW-Dr-c5CKL-cjShE_Qs@wnGa=A^PWcpWoY{Exn99EmZ8q&}TVzzP-CwhtN9-J0M*R-MSlO
zj5E&hD_5>S3-a)x51t~9H7v)O`J^vU2O`WciwshAU)xHww~!%oGf+?`1b-3OJzpA&H
zoII0x;NiiG$2By9JzgR^`(*)XalppSS#7L0;`5YR>-Zf`0JVVUfVQwW7+aipxDxia
z^Byzi6%@FE9)K!^*fx22KTL*^%A7lwgU*)an(EYO8Q>`J^$o&JLJgHks_nWOkEK+R
zpw`o2{XoZy2ko$wc}fUJ7p`y8q#htJrM+Oo#QGRF|M;n&S~qt{Yd=o|ua?2+xLY;G
zd^dZ2Lqo;Vw1$g1ZYHq%jtHBj$ZOZG;UNHr;`J3#b#ma{QaUTPEwhDoMiC>gQ5S?Z
zNOT;@ug;DdWvQIX`UVh&^2SgMXg=B>Zx3V$Q$U65S(Km*17cL+Ga~bWpmD+ow7<8P
zL2W7$8t&-F*N3RsVA#^I0!1DM?9gadCHS^Z28)<6YvvELL(hD5-(g
zFr?qsUr<;GCZ+Jq8+&rU9_{P-`BQ+13kR(PBvAUt!aE0iVNr}@lbo7G_q2k7#X9w$
zu|3-USz)1J@nitQ67%+Z%YVLLD@UnSaZ)GQsJz2*kP%fWXq=<#lzbdJ1a%g1qTD*I`Pk+o`P&9gG*J~77gOc{3m)|B5E@Ds`071K>stgxR8|Vc4oHVQ27265pz1HrVKhS;xDKGbY
zPfls+U#Rg+*$Qs0lQp{HqsOLE$lAfiX5Y@_C*xnB5{M!wx)v@n1s13%qj#dnS_Kpz
zG}j`$kMPVf6^76&gVHSs*Q?LUEMiOv&t$;WV{2^tT5$Ups>V+b7+v`BVY{aVi}v2r
zn{k!MAwEN(>~cy?trq<%tzTX|d1Pd4EMi`04+aY5WGcbJ^L@T3{>k-K7l?ct+RI)D
z4PbFVPHg7(;8*8O&CIZvv~s<$V2Q*m*K8XzbFhjs=NtC*SmeDh$uoPP5vt1wi^2V$
z(6cg0cW?PkcYMmFdZ3_
zBbct`gF_T7g4}1ZR=o#HZHEgr2FAlwA-tN5wWg$|HYkp~=|}|Z<0uVEMA8_p6R=4X
zOXTeMHIgpEQc^Mh+N-KkPQ3D}7cdCB4p_NzS{*bJQk)V#&{QR(p0tP_ptiv~yn@n!
zG7Lfv*@Fi|LPE}WJ|m<>-3|U3CB>&7=54yP{MVBwGaQRlu*b+yu&|sWodO6H)kcws
zkgu(gh(s^W_dV+73mWL}pM=d{FDM^qX5{XNyfwObVcI05EwB2KJ$LvSoj>DU27z9<
z$jt300UD@Grh>ZH{ZswfClh@oAMu?uGgCjDO)!5W`(E=I6i;oE4ufzSiA(*7nm0K9
z-drOWuu;>}qNS&=@EnUNpn|R&{EV-!F9JAnHv9>^)4SnAs|q8`F9)1NAje@A&(3^6
z_&0;`40jrMP`|PuvMLVM_
zf&_u{WzLnav^hLja@MUt?cP^J=LWiR#`3008y_E~cC5aSDk8-MxOH${?g?o3L&T)@
zTf9@95#T#Q9Cs2NOIvA|&Tb4?S(?#t?Da$rX4q;%e95L~*o33%#Og(nqn;gM4JY5w
z({0I5pY|boqcuyMpFab5C(9%%g^@OEGpFtVRH#x8%_`P*v@G^XF~iTnXvAf~e>deg
z@I>oGcJdY|*T5*RT)^gwzkmM{N~#poRi7lC+H(NCXBu!LpbTLFr1?v)o?LqC?rI+-
zj;~^z%WbY+y_#WUZge#y17cETpMT6uOx#Vntm{!uHx(uiJUjswAUh-;{aG1%TA|Me
zd8ELoAV*VWIq3LDhp$M@+XDcDii|EY*3%R97}gz?@2RR9B8(tqE+f)XcHBzkxs4X_0DFR=`R&Zl&zX$0X;-BskQ#tOx3RZ}S*d^p
z&hswN#>Dzv`qv(bWiXZa_2Y*Rs6k6z_@?qu=bm+&@`I%!>r6>4K)?`zON-N~XS5)d
z(&J-48=)w`hR;!CcDG;{i-U&K%O@cm{5LH{a+
z7{l80PGs+TL7Xz%q|rSFiRYKz-k_?_p^=e4KeciTv}`~Od*W4-(?ZUUs*}cY{yonc
zuBJ*qGKZT;^C6)4H#=6dOpef9j^hYtm0Ta&ovo~_^Y2-$bc1n@;jgkfSTfJfZcC4d
z(876%T)XoQ$bI|lDU*#ynkrXU1sNpGRfJf^uC9JEVP5Bobc2tae~gmKp1Tf>t#|k|
zGJ0i=;d^Sww@g>_Nj&(GYHMxye
z>^`FF+0hs7xNmrZ7mz(`R^QxiFc?S>oOeRU8
z(Dm{>*L*IL{BS;=jJ*~}jr{!l0-~XSL~$@^v)6d1=n7SAy*@LT97A=qT1}Rlf(>As
zvs#+a7)yl1J5U^=V2$txRN}xnSNf=qegYm^D>&gzz=iR`zfCTVihsfZXTV2P-I5+I
zEO?Jv4QA*2JOPR7@ms2N8|{SzmMj>N1e
zWbj*LW6j-@p$0>zxSU+PqceoWSv}k3ugZAVBM`NgqA$v~zj8xX6fQ11L+9{LlzJ4+
zwC-!F3QxAD4k@1~;i6ee5P1Puan0JjaYJbD-pL9*(})9ov9L
zYaA((GI~XeGd~St_eH5mo&TCMdutOUcT2R#(Rk^>94R)-vG<4~@BrW4x>o0Olo|{V
z0zi6hedY=l=7q7+x8Eu5e}?rucR2KwA8rdhv&tGT<@^_*iB|h+uJt;prw;-gPkm_`
z+c>lB6Nnw0V-0JG0od7~2O4&ASjnhqBD_`$13yi`cQvh)^A}k_5CcjPnDfoE;n_}T
zAQQ)uZ3I-z%(ljLRyE&SH>xrp*Tf5$DTtkaA@iCL2oO_&(HAniKPe^L`gVQ
zi-WnYPC{7}FsVA)jpEo##I|2^G5VQ*Aa2roy}|K|tl!F^gYzK-OY|mi^B6B1^Qi16
zDd%q8Be{F|_?-L7Pvoj1+4r3uQg>0$HqIpq$iWNj{)&MVdVEmuJlXRigJtVh#a!Lp
zEH^T@t0o;?FX8j2p40vZDmF8f%Fq;r;r$g$O3Zl{FFzqksvR+WoD7N*D$I0WGO+7f
z5#kYb{9w9dt#&hce$;?-Tz
z$H!#ZRzl-|Ia=Lmlv@2B5CY>^06K;K5&cXL^obb`z+yBjxR(7o?uQ3NMiy@P!cfms
z9a@HP3Cv_V6-*4aEREQC0eTtzoY22FZ=3nBaiPO676i!
z7bs*1^wwqvk^fvK|XUSj3-d6TJci=rP`T
z0nre>zm-S|Cxi-)Zg1}GGeAT+
zd+Jp3x18bOVH_x#Y~@i23EAhUC}u|>V@C=t=v4FJ!%Zt0=)wH`{iVNpE5mYggvb#h
zLqm1F4R=N93aaQ5+&r|e?k35xb+(W`Az_~a@~5GtT}($~;yCaf&yS|@ye|OAZ$R>W-2+J
zG8B1OmHG5(1&(6uRo9D!AY+En&xX53$mpTz2&R361gYnI#rGpYkQ3ZyuWs&q4aqJr
z>k+SUwwJCzTtt$L;wkelA9!*XT^OuJoQbJO|A52-AbRY8Wus>rQF2cjc>KbYo{}<&
zWp#HD3SD-{U2(Ht`K|h1y}AUBnnlPo$#sjmQs~pC=e4zOm7dD*rn4-+T2iTsIrI%j
zzSPv5#Tf?$L~lIm@(^ec;`8Ve49oxQyb
zo+29DyqwFBi39v;K9FgutyMc}y|S_*>AR#r5=B-;_ap$3#
zzvtXMFgSQOHg@kMeX{O}NPBB&8-Qx{-dyqahK8iWv%~GEV4S0#y?sW3FUq8CSx>WN
zzPl}&Sox$*3ZZI_^oUUO10f#Q)GR}d9ojk6a6a|TM-kD>d^xl5*p1!nLvD*u(qo{;
zVuv7O7#t)co4ye{X_LPI9A}u{JOH8h
z=ZRaZJEn1r8y%~)+V-dcsRvGSuw)$_9Y48d=jYRZSo=ZrsU#oM{rNSzC^U3*SQkum
z(|j3e-AYjU@Tm0P(8>q;(CYQ&qP$lp5hI0(h^u*!;e#AY=H7RwGm)Xw7%vc)%ffZ<#kQ
zz95R6pf=y8y9ApTInJVJC3w(WSJxHgLEHya!QD2W7e{J`;v~^;W^+Jlb`bH`B$G-Y
z=8Ub7S%J!XVxl*G_OS0#F-UMZ3CGq9jaU)A%n0^ZrwP7piF_jX3Zf#>FxN+w{{}U3
zdGxjx9aG1J0_6ZRjunofd@ko~W#PdFE$kaSTxOB+UAlOB&`xANw6WqtTJ`n$*~scE
zugb7wiEBwLQFo~J!M_P{M
zdZDL87LuSMZD_Lr)&?M*hm=>9Qcvo|H3p@KWBut4j#sCDx1@iD(yEQ>b4!am`dV<$
z4oFF*U1o~2$N%064JFzS9eZB{_RRv9puOd+jm_isFuA=^Y9LH?d|DBEH&W1aVv&5A
zO+X{X#R#V;mqAc_7;u2zyTDTVcW5Nhh!MifkDH@LCMP#++)P(?VUVI|%h3o#S!kSC
znVC~vr^)RT^Fj=<`AHJGtr0>OZC{~EdBo2*Li*1`IZvXE4_zfr+qd|YUyq8?gLIBU
z3k}nFz*Lu*Agy~K?;|JqX%S}@)$^Ue(}u@TGogEV4NtU^5?pN1xa>~|$Dpo5|KX1u
z)S}#ph5%y|lkC|#B6o%MxL)d+HIOQLzYykfrQFLFN1aDEb%3hL+$sw0%(LTRsk%s!
z{e{t>yl*Mc>wS83IH7`{CJ}0GME23M$+VshKBJp3zy|#TxZCZLzVe{CaHuq$>yB57
zlj8|e6Nz@Wh7Any(~>y$*12ADAxe@fh5AQ_~CtLl+m9GiQE`jBE$@#l+N#TqgLSqFQP*
z5F+L>0PEG+ih-%MT6+oKYR`kkmG9C5WaubMqxBfFD>NSxD!cJ@I1EQjYb{P{_z%&J3yHP^GWB@;}Q%?zyw5B4$T$_cav06SqmWM@sUVO{rek}e%-V7SAEmh
zcLZ`5ShbhZZTjE4^!OF{*m2e_FD-d6(VPbtt=Ga`8vFn
zCH=2|?(U8^d#Z@$xXNtOdf@{@Iuy}n{F0ZSU(MtVsg)iL(4ZeiJHf>vbgBm=Vp5mNbxJ$xlr2gY4VaBZhrzCrKUJnTK}(y9
z6qG^lBM0$hhH6sb$y%o#eu6y)!*hn^HhFWVrl*5(8N-*z*iaNo&mXkj!RUVr?Iy>L
z)nNaf*i1%}baiuU_N0thA9*Dy1BE6c?iL;t;kWJ4)IDJ)65EG{1axj9#7Ik<@G^FF
zc6xgg^}%--4aLhirK7H+lPKYEha%w8l`E^0@953L6pL0*|1KMhSNnQIUVT_igOW3)
z8#xiR?^dDE$EME$zB7$Ka00o|
zGenir_N+)O6ndW^fQi#@z_=p+h^E?yD1!32Z+`DeCBWpyZzsAxVu!Qr&O0`aEiYxeaw
zujh2gdd@Az#jS{na>^RFB#k5tC&;lS$cc@#U-uB)eZ43-Dwrpjs%1Y910T_voNr1N
zkc8W@eme|EUt_+^F^s`BC#RM$g;V8)Wt<)ae|SSY2zj`-1yru
zM-8ZGqn8AN|5B6h(x!{O`%ddh+%2i*RtJi-zcfzDrO{izbT_sJIE>n#vv*NTeZwWf
z5PAshVBmIt0Mny=GVK=m^vt*M@iNFgwu_(9&HUKpBD-(j2v9rkK?tKTZ-!PD$L~*&
zDiRXq=a-+qd;q?tTCwE9`8i_LO
z+rPg8P@>4GLqJeay;|uQ>#zUO0;El|tv#uLLX~)Z`^NT;T~drMkJYl$s1)_JZ6=jJ
zj*yVd3b?-^OTIDpCY$l~J#?C$!lp}%WE&X6cx7Y=@QA_|p+R+XbA04}NG%(Q$4{K_
z$bU*N&pLW|6vo8d^J9v?4YClxb|_9MS#7ktf|Zjqf@64WEF>r>>ABM;Vc}e)Vvt`y
z%gLb`$?4tnj|gK=tu~X;#=JD;qZVILGRBe%9-$Q(XNe6Q-M
z?!usCf0gn$?y=a158PFhMD=_yK0~x8PF5^`bYsh{;Qk7S2V#9(
zA&rB%@8b7Zd=5lpA08b>Lw1W6=O;f5Tf(W#Be)2<7i~w_)oy>GgDvRl%chIqkkz-c
zO6;rwYunTF59uR`1ku5n=YYt};aSi~1FAEI1szXCL$RTW$x9xaVzl3)^q~}xd58i#
z1ItG^K8=a}!QZZyTpC%~L*KPy1r%x%#iN;{W0|8*w{4~(XNwyZKQ~OXWB2ult?#b8
zoI73hOzPU`#+0Hv2??(tb%q`Qibp9=AqLDdxQX)yUkk}}!oiDq2?-VkAu|u5S4D8#
zNP?aa=OhXML-T%9-^eeIgZIu(2npFN_J)E4*stHTWe%0S+ufUyb|9s;zr!}v!eR5@
zh8&VeZA?m2oS~kPm70H$VXSgBMk$%b?7;0)&Q<&Q1%-tMS?@)&+>3cu!hGmbG$)f0
z7cXOF!~XBzW6vBqd#a~swrIFrl_q;t$|_D)boOS??UMbJDYEY9k4LQohiDqzCbGpy
zKun^2zbb6oM-_WDZS8We33if#vhrz^T>zXsAKi(M59c^=UlUZL*%zU;p#{a3BBXw-
z6co#NxQtM!u(8=iO2W5BM7J+yL(|nwCxXpWztW)U&AGnNCH&%x&jbW!PvivN-M?}n
zFm_crChbLe%uf#K5P9bfq&;)ho#~DOo0hKW=lu2OXy`hat@2CR$ieg1zw;%(D+V6F
z&7{ZP-mS4a?h2iQ;vrBC08WkNJDYQo^fQ6lpjtgYs(`ilZh*T7{f%IXu#ouNS-z#f
zp}@oAUSCUH-IkITO!k`ALum;UKqyHH$+{?1rnp?zH=@B2RLmkNSx~{k{@3%rXRyou
z@3+ufJ`*wTg|GUpZX08MRhD8YM1dI3@vl3}K`m0tz~o)kSGIKd&`po2P9EKoP*XRs
zoG#B4-H`buRbW<8%OK*Eoqf3QGD%CC=Of$B$QoO*1UV|oP4su`hNkZIs;3i8?|`0B
zROAq{&CJd1aW{r9p1S%Lw{r;F72@6K2v&ffxxIu~Y%yvZ`iGl9432V?Z?(y|`-9=X
znD|4Xg&Kt}N!s}57x$k1txfWOz?DA5o9xi|N9)h5)}IN>bB(I`x|W~z{rnyq%MugA
zaxeN>$pem)W*16KbB@m{n(h{=R&bu=6hVK@TdJF5w!f>IPg@-==NK{#Jhc-qv3Ghu$Q5WKXoJq(n_Nz2=qEw9cNv
zziG)F4xFs;^}*tYa0Vn33lr1u=%~4=X>n<(cHDtCkkT55Z%X`ptl$Yip~%TH{-Q`-
z{xXF5U}{#N&S)Y87z~0C7#eDYxE#P8(?&E7XB%BByGiK13vAU5YNo~Y&CN>zWo;Et
zW~B#mJe#qd7F_i2-E>e;a17ifiNwkI5NxufIt{M_ou~u{%v?~qMr-P&`Clh6z$E0|
zz|11#i#Wh(C<3+9(I7ZK)`4*b|EwU=u4Q5fCSr67^GwSYTK#1G)C~ImRcalCie8SL
zkL4$)7p{$(%PQ}SU*Tp^*W;>OWql;!c|bMm$blT12U?+WL;eKD0f`eQh7O{!O5*`CS3)okeo%AieX
zoj_#fpxaDpM5+RXN3izO+~wbW#Q2!_l$0jO^9eKw|@g-4*U=&}`k73O#u&xB@bgy+k*7sRUk%X^9XvL$%A7
z|AO4oekd1wZ-7_F$fylDJeG?bRK@s6Qws}e-Jtb*W0BODbilTCbJg56EB@c<>Vfet
z&ACaJcv8x-gk8WjK%o!2A8_IOZ
zG*6?&l~?q3x@=VvBLv^L#TP*>t86ifM<0b3dV>1j1iHA$?nG|`q*{pl
zbRigIr1Yd|*hB82HcCre7`+J)M|6#8j&dn;(^4_98wLusX!$BSsMFDB%S+rENi&>%
zQ`<~gN~@r@i-8jLjq4dS#9%oj{-Hh)k6jsU-G@ht=LvZ{=ryD>cR%!FBoMMbRHY$8
z#B?m4ycS1$dm-~eHGZZQkd#Wvt`J2by`C-KtG3Y%iyXU1jhB%s7;^CbA3s#nThOvZ
zG+QG>M;sx>ltJ9_aVFge1t~RNG7>uKON)#7MajIbMJ%<|Zs^%?WmgpP>#8c1A#bo8
zT8x28SRJFEu^7NS)&HC^5@V07Atnyqc#r{Yk$olIa?-HMs%9S*-M}-3AIECYT3BgcSy+39aenn741Q8Vx`305-Wj;?jRAcHO
zU*15%WV0bK+?zKaQ&l~K-t6k?_SV+QZ;l`DBUR=Rcn4zZ+3yi-0kkY5#$5)dhtWSj
zN~kGLm{*{oFOOpjh*24dh9o=*lo>XX4zSS_DnHydj>&36{z}Q*ZTfao{ckqfZoCF&
z2?+W-eUE}DzlwWkvOE(jEB?ab(^>qS42B4S;B0DYYHWOl0}L?;i4*7uTxwYB7K*pX
z(7?V(0sUOx!hwLp2bL*O{4Cw#VPhf$Vt@)nb8h%vKPrkC$n@B8+oU9Be+Tb%EK@#jh_!~$W
z+1bCpyfX3Cp$bO#c`stJ)~+ZFYFh>!i5ocCuSoL8a7?5sL-T)7mw0wc8T~;ZLWv$N
z*~1uOhEuFHS$T5``w}P=VB;#Ad=F30gPx;9zkhQvF{wOh3f*grYC>Pdkv)5Updt?%
z3UnrJB=K$JJV%4ZGSkydFh&ttEi0=&usOebOJ;x_Ip}KI+c6#LD^v$j0tIL{Ao{_P
z9$7!{A`7U*oGkRuUKv<7HKfgdvj50LuT=L}SBL%n_H
z5npKcFp8`~z}!he7HU3J%8-$u8$esOxxF1sXv)KfoUsjB9#w
z#UK*VH-anS7ZFL*<)@)|hb{K*-Mg-?MO@_YzTpuJ9YdSg9mxfpbI^_m&F+)Pe!L>u
zUwP|xY;1@`OiWDqrI~6BbZ-9_8j*~f!NW6vAB7O-&`uo#gBJYmrlt!B=+LOx+Sp4F
zOAC!WnLT_+(AhXRdaiQ$U9&y|u7@bV4;u?TEo!-n1rPCMm6
zy+3+mq|S9Jq_@z336
zm(l+v@zs*H*fIRV6__-dp3q(oSwp1b>6O-G58yVmdB30_ns7Yu;PKH2Ya~{Y
zq_ODj+&y*@83{y&+p$7fSk72m`)hv9Fcb(@+K$2jpb#eKo-i^B&VEer35Eet@#BpO
zSCDsTX*I;DrE>s_YxB$A3c8iO4pgjQsg2IMwu})k^)8q10TWG5FOGf#viMy8JL)Y
zPV}%+ZbEp*y1~k_N6>%wnl9W8`8axSw>pbK3rM%~P)0`Lt=F%YQ9Om1-W%GBUITO)
z0ZllC{x$;fwa|-!Nzi)pzCgbv07*3@6-`Z1+UYmmW0>z_NKy$AC7w{#ac5MgQ3_t_
zTV3hk*Hh+>*v!OkzB9g*orP0VST4?HNB#>|zH=gbLux+%>@ci1VKPq~XNK|%I7OlYue-*5{`ircgr+W38YwVB
z<;4W_JW!089)*Me+G
z*Gj;EGXK|A$R_|mO7J`i-+%dXSV(A`XkqyLc@<$uu1nQXzN@DP?9u6TTOfDzW*$S^
z_SLIp9LXEDs@z^R;AcAb12AMT666c|_kVuk{G+|~Q{oYN(lG9ca#>k{X#rR?J`LN$
z4TDpNo11JK7_{g@A7k{3kl?C9Qkv(ivKPueS<%~lKgQ*3!^~@yze1zhOO+_
zm9A*-AXOq+(u>~k2IE(eMt~;-C*6N$_GMq>z}}3gNh5CY+C3T2SrY!Y=(#QGQpHJ;YpU-D-Fl
zggl{vQ{$UY;dZAGha4Qj)J{-BG!FEl7?g!6KLD3Di9{i+s7Uf}V`Zf^<_rwKLxbJW
z%oY-xjH@-aRencSZtiQ44|jqIsh`n%A~a%+;md620d0yz2bF<{>y|Hpz%U5z&>@cb
zhR&~FPnwvp3ja|{)yE_U)M^~ho~;?2!j-Nr{^S}z!-p;@8=44Jv|++-0x0W6lu|O0
zK0pg8zuSQS9^yF?YfylvdT^@1r#re+oIF4wo*C;IIT>}{6agfLbqc$otTst76DZN_#+fRaNI{LaD#6I^Gh*}MZ=%(N2Ix!291(9`jEaER7F!6
zN+=o-2CGdUTz7f4*WHaocz^ozH#9@%dy9FPsG1_SsL5@NKAt;sLV>%)``UP&$*xa0
zjIbE1YY^3OJasbqDFJ4FDRB?c1keLFeN%go81A!cSL18uxw*MtzkX4(2r2BHlwsY1
z6cIVvWBRCv&V{3bF7V(P!EFjAmtBZ^Ik-pF_XJo|aa!T823
zJmuZAYcs#Zl82fR7emWOMybfV@BZod@C)p9;Z@eVKa!RfFK_Uf=|Znc+h%HID&~56
zEeg5pfhliHcB7Rj&N?@G0dnw@j#ogbkVrQ9GJ=8)VE5E5B_t&=pZjiiA5aD0Y3O}2
zmR@bDY0BWY$fFP#2^SMNIr9=lU91&2jG_g38zUpe-CmQ}^}GdD^}0UlMoOdut0-k+
zZyxjD^YdFpY>{+UJB|)4uI<~^%a6>pV6Yejs1P(R=>l6}{`-!dksZ8qF@S1EUxxAMX0}2Ps|i~Xqu4kJ2Yod)$~-d4xj7yN
zHI1OpKcA4NGoXmvq%ZH;j64-}-Tvk@^varFcx_wzc?%OeH_2L4dctz0=gLj-FDl26
z++&W-?(}%B8p`xY^U8u1yKK8hOVyjs)~LE?QW6rO;Jd$7~0%AabMa(k>6`_qR6wS2R$cMbQA$LL--Ai3Y`wXHHw7@QW8#
z1*Tj%3zyr_-FR>Fyi5NgEqaUU1nkh)oexA5rh+w%5E3pJZNAF(Z1+u^b&kPgkWfKA
zjZ=}~7{8#Pax@*jpvsv$T!Cu5Fm*5+JXmM-%8NFWF93mCSilgQ2nP)6GND^(>H)@V
zXK?}wUjr;3$Px!|6S$WM^_4WTj2tR6Q`5M+cUyui>bM!AHwIILYBGs_8KNQ=b9y2D
zY-EqY32NV)prlu!wdZMnsz+B__K*gpQ$;#tM`-!7`uPl?4#WSDKvgsA+#$cPC(yq=
zRaq=Gzw6613Qsj6FYj16gA|QN%2hB)O-gDRAu0?2==cZo#?kAn<;IZ-u{9>wVb0Ac
z2pQD%fh5(sWDx^0cixRqc5@?yKT#4_FtrQjJs%Q}aQ(c|q%Q-@1bvlv-Q9Bh{KM!t
zi0WGf8Xdj>;hDVeQYgn@77d2Zk)Jr~2T68o-#ru<6pzwAXhdOu11}#c?G-q~1R^&a
zyoMC*j|F+)p-LAq@In9j^=hz;T9-X
zqtnDB5A4v=%~Hz2f}z5_FR8Wd{!&mLXD?JA5*s{YXxE@opiBGmJGyG`OqAJEhV*vP
z(b0iu%GDv7k?1qp4F0rJTv0RL4v~dAcN0`pcu}yMHxyssYl^>c4A593=9NuF{+eu7
zeEby-u6d?7=OqNH@UktIT*Y^UW+*HbE-s((xUYd6ZB*MhvhgG*4og5`?DYjOWw`#h
z_k0H9slprhY!r+PO-W6C+;CNaW{Q+yW$7P+MAGerq><3@DwD(zb;6S{pRWWdEd&DVx-tewCm?Iz-
zcisLS&b|0SPLd9SUl|xYp#%PPV#hJGt)ZoX!U(Qo=Iz^SU_r3qm9SJy8%?6eYYM^VPX!7D7+2;
z-tLs=ufQ?9cma>XC$t1`NW(8*jYcgX(2sMH5N>|4XN5Mhxh-sks_qXt3DuOY^V$%!
zIprnOV3}JdDM>o|`e)H|!C4FM=m)V_6Jazz37rdKc0SF_G+@Tiusx?38lTy+=iEWV
zsB0K-pdZsTN~HLpd;&*<;@24c_J*6ZJwi`D#cUwK@mo5@1;}4_k_$)cJT(oOlf6Rj
zhCDX!X_*G^otevlOgDs2ovoKXtm6M6;Nze{NsqU_a6+D3;JWDHr3b%)lNp-Yx?ZP1
z@xblELA4#qA>5d-&}6){pBt*C`Mo!;6})fw`c#{emX;ezQ-BpgA)&}?*ODGx4PU#u
z^N=&si@$5YCm~csh^LGmgtOn`w^NvIv(YdRyhpe?pPsQXaTHc;*;KZc>4hZ5`(Wlk
zRzt%t&PKcRGTKL^cpKaj_n~@#MoWcd8x=Px>fuz$KGNAxfL6sS}pp^~+Oyq5APLUWRPo6=eRF
zm$yt^%4iz>FtEaK+7jCiVoWUIM2)nupGy$GX=rPg>5GE;!b#e6h{&86XD=0mcZPbY
z^n+flPV6Y8n;@Q!d29Z;@IP9Bd5+{mP0-n<}!$R98t&CFXJP1X4l(5B`k7eFgU?Z^C`k`hIhpP%ozg!>A@To9pxR$>P~-sn}=
zCV1={kiQWAd#~YcT;)<4dANl@MWK8`f_?4}Nx0Q0Q`<*iWX?K*ZWOL!yEoG9
zlQ_lD`kTM|hsNm!eFz0VN26}-nWXKd-O5}lIb9?tSF@|r0wdIO7)08QBO50_?>#lJ
zwz&KJkuFoWuM2ku4kk!9g+xU$KtMT)p)Vpj`X#z(|HiA$e)@!ocQF9a9|xWw&teYb
z;rRL9Lan-<-(IbV`>Apy{^qVFb6wqTY`3_QMTjxtpMh4xg@~Xaa!ShIQ{5X==^kN$
zj;qb$Eo)e@Q09r+Iyj`Jq}Vu|L`x8lE4&2tH1i9_Uc4ilinKa|smaN6qfbK+kIYwp
z5q9+PlY%7Di<6T+omYuLgZWk9s2{0fL`NXj9`9R9NK4fxK&PS+)U9Eibxh9@gL9fk
zJz8C_5wfF90o*!(!UGBi`T{mVx1c
zhYt03buBK>-@Y@HiLWRf+fss27-wVnM7SLe(#kPcTf{?@FGq!j9z|OddYMP=BQmB%
zP_UENs+Azj2p|BaJUy9fIHaVdr23}6M`SE*YzZOEn%;kcR~P^kt{vNS*3%Q|y9zEL
zg3Y)pz(5PGe|}I?Gw|va-618TAOO$v7cRtqm%Umg1u8S}Y-0Q)x}8?pdcci`F3^55vK4(@&bP8`<#7Y&%{P-O;~x}id;79eF^p|B`FD-NCdMH2L#?T
z$S`n@vRT^e3$x~CW0yQ-maikK_&_8&=RzT(^$j@(wY${w?V6|kvX9T9ixv(VO8=*-ps!Z-6Bf}`WX-_R
z!DX})vn23QD#ds~hQ+cC|5&>8<%BO=G%*a|i{5?2%fldvAYW7p5maYEVh|0c3L8}~
zFERIV^y$!-Y#U1S!98ao6AYc-Pfyh+`{rPxL(*`o9xEUM>y~G@2`Zv-vSBCa8B(Ag
zfdvi}KDuOB%Ui<|Q&N1gzFm@-9}=*>_T$0jaPo`Ht55kv4L^99qg5QuF6ki@
zTv`f_UjZ65RHP?vcB8F=F3k=YoP8Wl;&Q+${}Qu(J=2Xo!H~RahH6$|~WSmM0s_6=%N{Q1&HDIz_a*
zn@a5(cfEqa-m(vuYae_dL?^o*<=o*uE=eO6Nh5aoHSO&}#a_v72NQOpZ)nyyJpl$A
zaa^d|5Eg|NsV6+jwx-NTfw9Ghh4X>$IzxkMy};rT_=HhyZ{|
zBse9J=9;v%PLAtXFSs~-=Uxp!nJ-w_7Rl}29{qAgzU$NnZ)4te8NJ_z+UV60Kyj!3
zUnPjX_m6srce>(|ljqT_Bn8IAbfVGDCW4Cz$5bdSyDxfhBn6l@00MwQHScU*Mv;OY
zd6Uj_Z$Az+L)NP5YQlp9=dQEUYWZ_5#L)Nl674$tuHBbsu=V#IbCd$UDC6hjQ!$J*
zUwxuAVk$g<%Pz(ZvoPSvL7?;0+gi~BhUUWC@dX@_kk~$mu4fH|V*oH8C;Gt`VF?p+
z)ld@Xy0PTq+=KiBIE0_G6sIMwsNp}iPjBLCe78y45kq4mu3b~%gvcs1Bn08u41bIk
zAo2a@;6b1`xBFqp#$_)2;7k7)`VlV8x(=3c&JZXeESA~F4jce>1>3-_9QuJ#9EmA-
z&D{y_I^wd!KrYzK-v)e_(AZC}-n1cvLY*eKP~@ezzM#k?{c*`R7N<&uv?N1<<_a^W
z*KLkV_7gHuleaYbl&1@K25Y+TS3;R^Grr^tf#i^O_46k=*%i#60t7=86yjg58hGEU
zsX?ZEMxMw<)H*-Df3FAD2a+Z%d^$Qhm`7UF3sj4VR17ccE;Av%{alPErK=N*k5L-%f2!8_L8MmxGJY*A-lSN(A
zCi?v{vxh4J1Z=N{!nMkCHXk>;P3>s$VT$D87F4+4%;;yiq@^z)YzDu<7kbn3@(&(T
zSmmg5FU-#{`wE&?>tcQ`4qe;+giRz+S18jkW?)B=5RZ9cIP@kWva;-N^9l+KCn>Hx
z+a&p2L`X;}_WJP&+}8*t6xprp2WwmS%YEBVoOp$$W^+Wv!yD0j2tUM7r@ECO9K$dy
zOWP@WcJsV!$qR~OUwgYUswSKjgci{L*)00#x4O6(BRpoY
z@zHetv6?nJI}1KK8#UcnO@4)mu%Ng?Ii{B(`9)Pq%Na@vFE4iPxb0AeaFuSjl`cN7
zC@dl>P)zqfy~mOovHBn@obap3|U`cD^-tt0%w_^|S0$431JH6zSrA4@s#7b+)wgD6=^S(!Q6wfq@>UB@M!)=sjq2tA+|
z41N?PG<0rYHfY&`HNQRu1uQbvS-jA_SlDG;APv@37WiRm>NQ-~!@f6CtZFFqIH9qww-8Bf|KuSB
z(h+Jhz@Dw$+%27*F($GUWKQHLYI~gjYT-zM>C}pwGumH_=FMyrjTJ3UacbD(k|#WQ
zE?Ql%*SnXHj<UFF7s9E{z<~5nyc{Z|r*QGG5E{p?5mD)gh=`9fR#S{#X@D9$?!ve_Cc}p->k(_9XVjY&3E<%aC?Ykj1`%?
zKn1wU>hx%C9*yR?DaHla%JQE-=-zX@J&z`xDTEO;;&G`$V5Fn9Y86=lxNrn%v;_fC
z!G$4Sc8zEfgfs3tuvg_F?ZFfjoTFDmWwLCzOA}C)A+?OR+GqUZ!~`(`Pg+`yUcH^1
znl>(btCl5g)Yii7J1;M4i%uH+__?&QB4%}YkNl&gR1mj1X8Qd4{-n0;H*6o+-?Lgb
z?hI7guq?3rS5F;ZL7R1o&{NyI?mFa?yG$y{z~~uU4V`YWxP3u)B&F}4xuksd!abK{
zdREqm#K(8MTB=N%iPa0X{xZ
zF;md(srzmWQ9@+tD!R1D%DP!n67l6+T*Da~s3E+3kAJY(q~)AX|Q
z&YKBVAp_K33F_%!k#`jmfZ|G8Y(C8=$su@!@7vNPq2%=B7V}e;VtMVpi&`
zl=qa;Lf2@gtWnWMbzBnYp45b(*@vmw-T6f19h!%hKShOw`B>cmmp(Dka~~v2)=cLA
zIjTZ?6(Yty(@LhW#zt9fM~t@N_HD8o`*!|52DT&&E+YaLS`NC9yr7j#oLs0&Ox8lB
zsPYRyt5c~6=xIq7+@dQ8vGdYoC~oskESB4Uhc;R`nyHkeUiBfUN*KH79ugCOjiQHb
zrZ`a0$H>g1*uHoN;==OPEt@)3p1hK
zP)}c>(1gSFv&G}*N)|q;4gH+o@3izF&w;;V95~I$VVczxT)UL4Z2QF&LGd{6*U?BA
z+^me6nwT?}H#$8tBX+$5!||$+rcnqp`{FxjD~Br|F_N=|>xBqwgb0emrR)g-Nu*W$
z;@B}}HjiksE7n-lSe=npTgl>;q5MP{PBi-gPf)Tiy>Y`z!mgzcw?U(AlVSBrkeKaH
z@<%yIze|bT`lQ5h)4byCXn~QT*4NL%PbmY0A8@_4L&AHb{`I}q0Bb(l++6#@am2X~
z*Hx6<)Y$;J#SGKRubwW8ibFjDnCL})J=@mazHKgIJ9}z&Y!>7|4bKufzTFMjBJ_yH
z6l5}=C@Ho3vGxAKxn5ZG5
zi*^GW+)Ul5Cyk8e1p->0KVRp?&@=6Q2cC#Rf$D4O&{cvus8Q9;Ma*-2k`NdG&+MZp
z;tDT`GPrieaG=2RSjKJG2%QUNsf$A6z+Xmlr8N(2UtD-#aWH#?D2DaJHeM;-foq>j
zNI}^=%juZI(Ao#*n1dsrJ*B1ZSB`mAm=vz%TwT_&v`S1($xcm)b!)!vo-VUrYiBg=
zOWG1~A~1!HLrTZo+}xr5Pl~Gb-yeOoZuLzR6ljHnwAjo8W8&D?+8N_~er-is`6uO1
z?Tz|J4(_%q42rBoxh2zk3I5B_jytFw1Ww$zCT_Dkn&8N&KO$NC`_d)T5O&;eGVzHg
zAN>Tg*m?z~m%hA>Dp-iCYIV5l#;zg}S;XNb>-7?t!N+P29#7k!+evm^%7f-38veM)
zo850QvUwP#Q*%2=tifnG+&A*}ZN~APPmpW=fr%Q1QT}UJ7(Bj8Z=q0jE>X4l5t8>V
z1;ue4PtG`vI9yE~nEjXA&P}g>Fkh1nAe@^m$pV!cT1P;iL{0}#SGaZ)J=Jy$Y8Id0
z{QU;FP#7~F;Xp*O=IKM^@Dsi{^hIKJvcKwMC3D&Y`r?W>Q;(LU>evs)ylDYrL0MIc
z>cR~1ojZ4>p2MJFC}RaL9gCZ|f~Fr}b7P~QK4qR5#=-!t`wIt@lg(#-KAJ03C#t@u
z_4vnb4nGj-CGJ#c9)Y<)N0l#uXu)k^(d`F4u0Hw$w@4wm{5x)#@s;v)pBFyNsh6Hn+wjwEnYY%
zAtB%k%aDk!BZ-9WG|B|PBI=MHU;}iaiP;X=>3rMxx#ah?zB>RnC-^ql4boa{pvm2_
zv{hx)Lkc7;gh6gbp^fzFN3Iasmtq|!e35&llKG_%-+|HLd+IA?26Iug&(dw-Actd!MHa&3)TnaAT
zK+pq-TpVzOU~Fpp?jOJxr;FFAKu#YOZ~{GmEDpU@UR*CB9S1!(&wyGT=h(%6-&t_X$8Msx)d#nh=wC#Ti2MQ96RQ|w`iym#*>
zSPG&!1i|Ykvfvei*F!^urGv0043sw21b|JzztkP2%2&PV!O8~(82wa_k86XeH)~IB
zfWnzM`d2S9{3q9s_?C9&Jb+6ani0eFlM>G^QYR~s!|||XRbp|`?VD~@n^Uh+u8sgIN)GJl
zRZPxqmLnw6=QvtV-hJkNa<{1XUjq6A#2-e7WT6@|GTL6I49$az%BBQ^YZ)u&rh8O%
zC&|XVYVW)_?3~!`2unNXA3R!}&aDB%&2Qhlx!%zkgL^Sk@ibV@{=Os&Wg$Vq$Nlw5
z>AwI;zK=Fhx*l8Is|bCNeu}}BcK<2xsY=yK6_7l9G_&2ahaiX$YM6DIlSGP8`GM9b
zY7}%ryWhMa6ts1B&QqxfLH@qGfrjoA%3P(~a*eSo3T$L)-Ppz_*Y+NMGxun<*Sx_FWb
zodakiRG1z`gPi*Dq?IF*)f<#QM(SPJRW^8rCl`_MV-)$9JE?0>tGP
z9(F(peMUqragv^-6RN
zD6SpsQ;LDV&wRj3Jil8#AYtj#9(4G%4F`2yPM5kADfjx&OpUmisNL?=b6%|Gf@lui
z60nEyA-z|LpG^=lRG3(p|9LMV2)7g(1C|4s+ekV_Mxjz-B-I{P17=(*7@lr{l`ZCi
zn+D=9pyAYTV^%TSRuCl>(eo^9pm&EcB&6gBl-S5CDBnjbBu*^IT3SWzE1^L4!(vh|
zQiG=6lsM`h6>gSEMX}iIK7V)9(oFx}1U>Qtrwim{*F0jl0W?sa#Pz=K==g$C3!)zM
zj_l#K!KD$w?vZZ=dElqfrZCM*H@M}QahZYW1v&jp0!KF^iV~xZD9+;s0uAOB`RKpX
zXdd%AFJDZj`pMDj4uxUwL#;nDly}V!&~JFW;lp}r)~yZn&YCdQ7RDTh560aN^@w=o
ze%H*Q%W?L{1~M{(b&Ga98z3b*o__P>&L6=Rp%;hD;ogBJ9qCZtL73;0H=2=?rl2DR
zWE?NDeRSa`#VW6%fuoMbmU)gVngvA51P2&KF(w|m43W-~C0!-7e&7nZrO9+1G3dc)
z%8OcY*hlH1!bYwmKD*7FkdKlmtt(V_0S&;KzF+ji>}o~5@cJ_~(Ccz?pOIF47NIOe
zc7^0n45X-OpAi9l8paepdK3Uvx02FR6o!a)!12rP5TydW70z}Tcx;)KLlo%4m71_B
z(}i|WseD2eOy*SGCGP0$!rL%UcVRHk%#-|g<1BgD%TMFqz)uRHkuX5@A2&14^ZxWwaV|L>x
zR>}Xfw{QDJ9jy*^v9k|HZLn`qi-9Qs7rlsw4XpyS?JYLgjujBA5jj}48LQZr6-8is
z1aBeLF8_nq&?g$&y6t&AxkC6Aw2R0sPJ^1Yn}4oV1hJ%sc9DjWBwrI22qb~!`#_)Q
zeHQqTXJ|9BYjPSu6l6p5?{Mv+W3w6)c~TFKJ3kENfE*qHrZIg(Eb`vN$p<;v*^2ilr&y1@@;Orvt{wFOj^lf(&w)jt
zQk!^{E&ADX_)>pMi)#0$lObL_Fd0R;yy)y)TbRwLub+|pf+j3=?XVlj0(jwW%X`j<
zI5^O5w^$`3;P2&J1A;Yq+sV~GT$PhUO5L1S+y%+Q6@78rXoU`w+-}Fn6dF}&NCwLa
zNV?z%w{PRy-70^v;5<-V#>p&aS%)N@-Yfln{smQKWQ4y$3
zAwR=CvgLpCR0>Djxi)QI$~1pP^Cnw5yLCThXmV~ua79Ck%~t2QiKfOz&XbnwRuQzT
zncm*#oFx^Oj-OF)FW}Ck0b^~~zp#-6F-`cCBr9fB_$|1q-rbvxNiY_y>Rjjcs$kA)
z#!G=Wtdt@*bJ17hW2WXnCkkZb>+PUdq32=n8+fG91dpi3L@vlQib4Hpc40w5g(Id6
zh@73nFsEi6DLm*x9K%?k5bHYF+FEBn^?%A;N`b};NcXI)(#$S}qszY%
zAI{&s_pb4efY;rcxjLc3JBo8Rs^(=UEq9bL!0a>gQqg|X-o3ZcJQ#($Od}#T_V>4M
zI#WHBAH3?(Qfj*3JHTr0U-%Ul!Rs)@PoYNz<3Z)}v5;4jcVWL7M8jXHB+IiKOA}u7
zAA=2$jfQXN=j41qQvGc{Fhct9i@S{raOoNpnPwSml*bZ{pI;*N-^EPYhNvXP02E=xz3+2_hI5TbWAwK
zsiQI7j%GO0GfR2}*+8a`E*~lL#7@UG5N{W$aRY@O?zBe%>o*peqK_u(&yfAr16WNs
z*#rLnWn0_R8rZ3tU%iSQpQZx#c;3%K436mgWE!CoRF{$`D82_!#|+bB%eqmQr5JEm
zn@xh!L!GU(mWRF*2<$rVP;3mO1X-J7cZV(3=Lwo^A^UxlP@hMpkPm)|GQ
zvclvYz<#BVOw7!qGs#@Rx4BfP@-L8xucAjs-I!B|eC5*pWgRmUf!)um6#FUR#4lJ_
zQBsfp!@Zu}8rFJXSuTBu{r9f-4@#7R#3^UyEOJz|o8e+1!ey+iV%&h#Oz;@7f|+PZ
z$B^s<9@+C&L>D?3rb-l^Gi6=JF!UW0y^xdmZyi>8cuEvQx=~??InHoD0l8D_z>z;Ph$WJdY(4spa>9C?y(!6ODO5i
zm4k&K6%bFv9UzlWe#+T-2pA1Mqr83SRv{K7Q}WCWrH`&74b=?|Y;Sw;wp96zk(fZx
zNT{qFXVi5$b$Ye_297BAQ6^`&YZ?;&N3-Y8#%1eM>P`!xQ7^NHW`@qTpH`8-m|E;|
z>%zW;z%m~BgLr{D=?kkN6atB+lXUQ&5<)?ABu0n8eVmC`_r?HkTH27`Kl%m+h(-Xm
zgqL$mdwnCuFz<>DbJH!Z8ACmT)A_0!1aO2297W}eOz`b*hz{9DU=!OZU-xa&#K+vU_S6F1ebMK;E4AfPE71HH(90L`&
zwCJ20oO)?b`*Aa+p1oZD;KR5_{e?^Yx9Eze?Ux^k5oZ3h!kA~F6BHg7@6cD6&oTS2
zTIsR-y9{sq!)hkzq(kYnv+t3@Mo;<#UQ8zW1^lA2`zci8F!7+q^qtd|7(6P-WcBb8
z@$G<=_~9p^!>>+yjFzs)o?(nT!033G*HTVygn;voMZFjg51HhXg!t*B~@F=AwXkLlCMKL2f0XklH(0kNXTgf0ZM5OPFobmY~mGf*M8x=jbR(#hFi<3aEs
zga$ZXbF}xS=qr&+C%-AaCwneelSNVOax0GO<<(_b7KE~ZFLb4Op+W)|0=38%Rl@)uvmcR5{1$cmjl%rAtra`BqPDb!3NR5^D3noia6xl*4Ndwifw<{
z*@+mVdPKhWIXH=%fm6?yKPtB=XJTmITTl$Sf2rv=Kp5ht1ryZ
z)3teYb=SS<>%7};%k7`tAzz=ZBwfT~(=qxZfjUr^b*Rf{r8YY8G?h(1Yf9M4o*lrc
zppLr2z>>fpM;+~fx@hw@0c&%3;vkz+{qYOG%vig{8m?5f{niSoWdJ5ww}e6~OlTQA
z4@9+qt$Cq(=I0TD4+#^VW
zHFg4TFhI)EIv*`IqpyCH)K%)gAqYyho
zj`|&aEWIdMZ+NY^xfKU!;egLZ@v4`NPl41&bjWL}s-RlIQou=#clH2c5wL>ctH9)N
zPk$XB=lfjtDK0*K6j8(|i_Qi=|2Y43#C#?L=;jcNi;yg$&Szs~-5l~1$Q|A!B9iuj
zUoC6H;4Is&!C8PPcVJyGP4q-kG#zj(+nI#(_y
zWVaRvj-2D6$=o;hS>Aqm;#9+)_`pP(R{@t6S0`D1oY?|DrvA(%voq6jEJ6_G`Gh3D!$kudJYpj6VAD
z`}x;yG-z|6-IR1OMS-hpaPawnK2EMqT+vJqmVr|~uN-d+yN15!*w_!0&!(?t&;T78
zf^8dmJBI355FCvS4K`o*M9Qvja3@ZD?v1tS8pQbh!u8ccdQB4!g*n$^=
zSK{Nlu@e!In+Bb6{iyuqzUREXobH*W0uYHGS0?|l2@P>xC|HR&-50I+dj)RUxA
zkKTkHF)SoxdTJ_edL8aZT248OfqMv5(PLTKKtc~j7J@VIzva>6pvGq3o`cX0ocMj%
z53Wm9bYMOJE)!!CV^~Fnr-4Bdk_`L$?Gvit9FdVpzjU2W{a6v?lDLvH_Mal_WH;m{
zhAIZEUoI9?FysGwJ<79ym)$(GQGCQ8Cg|E&UDogT)4%w|Y9wZ4bzks56j0-$2l7bH
zkLHYhPD4wZ{PX5Ph~&h@t2&)d_!kFkp}oi=l}Mgb*TuCx%5(l*az-}J*kxSF(8bhy
zj0Twq8`)gNjwM>lc=NB(2pIr1Q>B@yl+49knfbU9>Hrll_3H%P0-_Or4aT-BZ*W=bo&#*eSf&W!;br}53zfxzPfrIleg%@?*Fqi@>FjGk${VWxGNgz&ELl
zGW@{Rl#~G|G?B;3$-uxZfARrd7OxY9;c6mJR0vJf5_A$OD$UpikN~x2VniZ4f<DW!GN}7<>Wju!Gv7_hb-jS@K#ECpQl&fM&?4)fZ=2xj{h^c-@Q|3
zK}4V=(QoM!cU%^D|2ND4!A!{FQa-CJR9at52QVimGw&Z7_>d!*?2pgS1i7ro-CJD5
z`4GB3(!l(dr|76MUKsl-{v!_9GP7z;<&TfjbAYv=TvXXRiVDvLuKhMu9}JDe<+g!D
zv1La&WOe<|x;Vb(X%v^1`ordcg9=U&^o~t%DPj0kf!_3bEOsO`q%iRH03}XX+}kHF
z4^8V^KnFiqZejEYYdU~T9^6hCU5xQaK0at2+)?U1Qv$4wIt^(1AOZv=B~xEbx}f8G
z9riSY(xb^SJv-}m=1iFtH5-T$APd#y5zJ~76^(uV
z5y9TXM8wyxaWc3Aj?NDCw~rj*c1Np}ilzyDD$!J#Y!$KnVzeCrlj3kG}K}AR+h(5_@a@#u0LgJL>rJsj45`Fo4q;
zB|IXZ9=7f$d}MhIC!Xar53AmqE}UMkKKI^J*V2je@!Wpz0wmJWN1D*nH@As|?ugL6
zcBwp1K<~KC`RHfqb-V*lND_~(vktw|S@-M8XN(P3)&sC!A-
z|J1$k1a(2PQdyYlYlT+GR>=lQC7okDrxw%*J1b1QV=eZYpj^WYK$(K)IoZLg4
zt&q5m1rbt>*Cx}2B!Io15q_AiuGeGE9Hzpe3c;gFgiqYF3zbIDP(mvG?ygi-n=<*E9CT>nJ@t
zdp(W)5$hkxtpW4xN1?5Lci+^^jtV#!IbZ(%N80D1LWpj!vI#
zPWQbmul7x=>^ve>K7Rf-GlSWg3rvUX-@QM+k9nx=MlkkDYGR_YZc<{RKF~)7yFW$`
zn=i;f_gdX<6|m|vO`nxlj5^dVYzjFEDPM1(Fz?2>7ybJR&4~?`9J-GC#l#?q>^6d_
zL%~i^oJ8^#RE>F1EPUsd@SThOT^&8zejI+?O)m%DjdT{qR##llyjf!U_|aQX7WJ!7s+uLbAA1%X%MW1W245`6bxnQojV{s1S8uz
zIx>O##~x@x8m_HfDCmp>1c=KiR?cH|Pfir?LhG-eKX1-4BFw)5d=7~;tH08vrRcQp
znC$GusR_EduczM+gst7XH5rs%DRZ{*>zx<%#g&zMG8Nf*#qgp4;i-O2n0*Y|u=8>O
zMkjd-7JR5#&;u~R0M3o7x`%g(3J4tiYJYX+-xpyD(j85!X@=z3739#(5>G7f=D0B@
zTRJ)ap?>~DK>_P`YZ_E^PL8zp#1!75(@e-*nTa)h`?TLtkp>}vz$*E>C!@jfplpmB
z?(gd}z|>)XHz%hTmn*&mfvLcJNAg!JkU155ueUsYtWa-An*Z}Sq$p72D=Hq9L)+N#
zl;P4N4RKy-o#J#q`8lidXHJW&aj#6OozWck@IJil>}eMH-T22;g(ufp
z(?#fpq2b}%LIFlg4l7y_#r>vqrJn4CUGcPLzcH%;7mhBI-KkR*@HX0=P=>5M?t#0r
zv(2Nr+*~{quTK*bg+)dBOf8c?-(UP(>O1@Q`HKC|^KIBdJm0wB=Qo8frkno~KS%3P
zC@3^}(MosW7{B=k)A>rr>~~Ij+S;9f9)gi$-_L`LsKREdAxDeijmvZiweS>6m9*yv
zaMe95==xWeTmRYLKlkrt^15=0ilDpCDec4)O(&jcU`@znHZRwgtf%#x(jMG%-=$q7
z*e09*sQP2T^KqN_#er^Fc*U%K&YfL<+Vn;Kti{T*U6-|Ou+7n60
zWG%ZU=pQ_MXtf;l{3GhwN(-1i?WP$i&^AKy;e!2tAG;Dj$ElHK2&-ZN1C|{dSx8;T
zfBwDo+t>5NAxbK40g`z3tS~PxD6bfbyM-t_7HA#8I#gKfx$TIgIY=-6eT
z5ri}Jm5Njai}lGjvzZNC#8E)%`-Iln&OCI%udcQMTfW&RoI3I-;8A(G&a&APX3BpD
zSeKm!Ujhnl47SmJyK_UJgfIqFUQA*{y?_g-p$akpO$`02KqDBLS&@10{h+BRLJN0D
zn+UWrh=bpys*BPRiO|*i=x*P-1)9UQw%qnU%9CFqQBlO;N#+ame@`sZR2!u1=sWwn
z{U##~mD|5$b33Og3K@8*|9a~yoS1SUk25F
z%wPxWe@~p>+|wF?EOvJN@x=LpKO9)rWFLqBW_$ecyWF2I_dwDRGn<)`fs~Sxz{2ot
zkD7L&$6<)cw)!cjD}dU8Erk6`zo`7X2=IqVmSMz^OiDjxXIG6TI$pvw9}-wOIZkx$
zu}Cqh#)J)de`Q~~gqcI9{ylvB0h#D9UIjd9adn<}Xj~`31j7LYf$ZU3uh4OCYT}iW
zN|U=dd7iWG8xzS=6HyC@m%)7H8d)-u8=B$9$Z-h{4z7`X>h$*;#4vMK>?h*(RkfbR
zmH79^aSO4>{{G|~`0w}Oqd!Bd?Y|$kn|Eio68|1phB!p{SP=3Gf^Y>rEjUaug%92b
zVPRp{wWh&ANkKu!=5Pi8=U(3c9LK!>{WWb2Q7!OJVD_672sT){eg*-qBM5N%Gt}01
zvBXcF3_uGAEgzuo{z2pK**Pf4%rNi$2=tx^swpya!BL=1AV>T#!4g>R_Ev(9E*8Nn
z&>FXvina1HQA4kL0QD6ZFhEgpNy*!Bs{mBLYCAhSi^XuD(Et0eT@n#0UI;(H2f;(d
zMpXIadqd8@!DI9(>PUn_?DZ4Xb)=%A(y3>EtO#OHLPFyHI_lX__W}eI;CoI36$EkJ
z@Cb^aImjs9p}UQkI@tA_TUu~wyW~&KrgpZqMGm?T{h?!$B0mPVGbZfNE-fvgN+Hs4
z3fYbyJH~$aya%TLA$=mK7BkJk!u0S8ihC0ixzRCb=By~oZ%3a7L&>oOnr#kzzlp*X
z(ch7I(Axp{G{ILNu<*v=%>D17TMIVoZ20#99}}KRe3>X~N-fb|l!RjiO&f&1o12*Q
z4GyM)Ye`9guvPQShv1@)6iA53$u_h{qpWTw@`VhFc3UK7
zWPAqjn6+|z0)Eq+y$HcJSsBw1azX#t?0O2;8&GO4vt_9C9xe>IbTr~beOBVlrmL^7
zyi$1X_`4VkCDpO8xHD_zu`@-RcrcIGO(1?oL|r&A#UwThR8Rd*Z+d!uASw#$Pvt&1
z0Q5D4C2e-MYVVPk4*&sxhmN1Ud!wk1s;F2r`nn<6w{(Pg*RC7L$woY8GY%n9daAla
zZC*KS0W{1nI>;@JCoX7HXCiWpc-<8uQV_$s_VO$ax!kq9f4&>(SE~yT9y|c1J6nIj
z^eYcF?eg;S7e8gvZ}8jTTo7+MWz+x)xb)f~
zA?LSn2c6O*CQ$F8_i>gl>uB@y=WyDpFAQMjJjPL29wejuZ!4@5`5egm=Yd)V47nXv
zhnC$Ewi2S~BR+s3!C+YPMo7tt&=Sz)eBHZpE1@0#{pAhfGgNtq4ifB5&QU&${*+@HU|$J}`}RaJ1w!{GUF
zOd?GZUrAFFaeP4FhcRXkmrWbOQLlnR>|Q^P%Mtdbr>ovdxp7OfI8Sf7$^t0!@!;J!a+Bihg`#J9B171aT>o6dvL%S(bl`q%grhx)?YK2MMxLU#}Nn
z${d#7R#h+1X|Sy(*7;-gVs7q{U8MyDA<`K#hYwqPZNbsXS9bsGWXvx$>|5+#iT(Sn
zm*{_C4`U%COPzTc{UM?Sf~|Njpi>Wha)MWO=2TCW@!>gqICHHYzX9R^)DjHYMcC(E
z)(x=fkCmZqQ!=m@9*&E$xc1CW6cu)K9A7?@S5o4HnQhPRd2G+gTKtTg|0ENJ(1%Ns5)$tk&w*LfFPU>gON6kO
zGYIe9o9K7oyELSoeitt2GWp<{g2V>#js8>lI3UCOOFy-69bIpd@I_l6mOEH2zzW6i
z3?jVe!O}zq7$Ros5KaCWJ%nS0N6RBz4uRAK+opk7!^05^?W~2Tnz=?$)&M|5T18vq
z)$!3)*R@cZ-yFW@nj7^O6{Elp-%tY$9yBd*d#iE!=l6L9!h>{>E+J^f`${DDIf(IB{>sSEv(>jHzoU5om>))x4g=P&_E7DWv7
zBjj&WBch@*q|X*sp7UXVHjQthC*a!?N*GallD#|LL%1Z{pKfzwRLNDP=lgVDh0kZB
zWxjYUM(l6e8af%uzsQs%;_T|mm$HhQ0oxoWA)_(djvZ(dA*!yjL(T*=W_y}6?5M9_
zJ^|Kw@O*Oz?JPWX*HZyMF$1fI7u+>^TZonssqj?J0QSU}yRx!$cA*_XEpRNXU}l_P
zztjj0U(Ck6{bM%E5DcD*ii)ywuy7AV|A52OIx-U{br5i8-O-LNt(FQgwF?uOFE$Sk
zUjpjE@D1EifI{fFKv;kp2tbPGE2t)@N+ytR^`fT&T`2%X2E|>ml4y;BdqKMtKhWi4
z2@(0vKr^_XJDSKO-G;DKS`rFqmcQJ4H5H|UL?pVMGPV%BC{L#~&0{VS9#@$KfslNA
zOI#sYK~h4(gXo9|T@UkZQ_V#Ei?LxmHV*<}T`*(?D~CCRQQ$P$+1Y#VPJ`<4@bJJi
z1h{?Ej&UQl=2ANh0r}D{UPGzfTw1*
zcXl|w2|ReXGeZ}*a6+|vA%YA&$4lsvR3@3_!e~goqI
zJYZ#^-^*zE(8mKWW5?`ybF$#TIoli-8e6kJ9D5fl$7QY{F7fv}zu1So
z)4<)^(bhdXohG{zS<1J{e;4TpVvsdYP)CnBWsyE;F&^J1Cim>4+t@V}9R5Q9EH8Fi
z%SuUEz##ilh?M>6WW5(Z1E24yIGjk2)|U4&;O!U|771Z$JUs-3-|!xIKx-3`8d>{=$dk&GDYt$S6dq2sz{I3l-`MzPd6`kMew%m$SA+^9
zvIyS~b`1GrDK(X#2B^~0(%@iefZ0E?3LyvzDrxUy^q7)w+(4zLy$OwLpAf%W
zGWNNn)q_^NJg5!5SPHPPe<6H;+WlHBEObZ!;W>}lO%us5gQ%!{wb@pWMvj7tM@@Dg
zyr;wvfs0Up#9WvN*Umqd+Aab|t?Ou*;BIX;3eLg#7|5?M7=#<#N}$hPe;TcV|o4n0G>
z-|OAo1-Ob^C{M31K9?8iqCa`-oEDTiC|b@qJ13|v|DJm_Txd|+1JhrW^lb>YZdD(W
zy9%mdAxiXN+#lZMC$vMFMRmX;@Y;9rB8_7F=h95o=B&;C%CyWT4r@{KnrrnX+)-GQ
zQQ`1(UAw^Q?_bSyw(9*Z7zdS9fW0Vb*;F=N&u+rjHe4L&gGVfUKa0E0HC_@@oIiC+
zQ{}
zPCs_5c2OM^GCct&QOlx4FVPO}dhrj%J(g}Jy?yq~9>yNc>bM&>v{lj?B_K%H|6Aot
z|Ls)02!keJanN4d-p*H7Rdt=HOmd!mOy!GOStY-|{`!UMAtB@pZ$LpJkqDkY1NEVn
zEbw$;oR@Rd9Ls
z&v!zA%e(nuEWXunP;fT4!;y<;Ewi!fAKrev4`)#`MKl@@x(iXJc?P57_jl4j%!#6Kj7Ba44ZAP^c~;{=60$Lfb?Sbqj0IgP-CN`nC2?M1U0$hlIK`Dq
zEQAp4gt!`32HWjyY1vJboR@OUjxOh@Ykn1>E3Fn~qQ=><#4&LOkjYPY7FQ+^`|RrqUSVO_72RQE2GL5yF*-k(xg+5|
zoR*ZN68i-mj@8Z&Qy;v=K&C*jXRx47gKjvw9HC)hg98JM@;2EvT~DS>Rz9Cx&pVSr
zjL#?|DeVsZx7BOuH@dRPY+a~mkOikYaL`Qr8dPp;7%vMHT
zaMBT1rBIVd^!;2mc0=y(3ZHkps2@Kg(bOhF)8IVB@`NfUqQ$1aVemTc9iWu-U5Mq%
z+J8!EQeAET!Gp3OopGHttv&iUW=9^}iwGb&D2fx@Q3bdhww2wb+0a6jQN_)fkqGfj6(Kob(kHpJ2g*lP&O^*l-(&dD2?uDI6ma
zcBY__gsL`XSv7JdVDco16iLj3R#x74B=2(WDQrxyB8~PeE*&c?fVMK%qvY^;I)f)r
z@c<)F+^hRBqb!XpXGkJW58TDCT6FEdlp^CC<|NJ5H
zKZI7|eYT2P-EVDgzo`Z>AzYxjtq&lS;o$);g!YIh2oLVpj~_jnGQNN!9MWFn@Tq~@
zMS}q&=1hcrJv`!Q=Ql%GC%~f_y&@}>wWZ=K-a^C$9ttE1{=FlL_Dc%r`1TC}c|ydi
zY8GeiNGRF#rjsh2`)?&AFxtmm;Qx8ndOI)EIoJr7&0)$_H42C30YDl*gd_=p+whse
z=8e!mYg3-@u-~GBhw+54lROFvF_PI~s4z0cq?L_ku8_Y?X)z+@AD+A_dAeJElj!;O!
z>lhvp0roD=&lBd^v-hbLV1+7tgwx!0kXHCmLX6a*M7`4iMpM831Htq*fwOaScfXwb
zNd0w_hAuK!nSVyD;;P=8tgQ$W8x)nz79?fmCM=!y2RBS8wX5~E>u&O7;#AdzL|3PD
zPh=iu*1f+lnU#Hcxdi%L6oy_WHqMU89#Y7FI2_{%rsI(+7TkHhbUiHxAhh?@RQUsu6QnPt$@a
z@|dg}k>mXQ6j0S}0f8P~hzVz0e{-F7bv+EJ7G_pJEHb7BvV<4~%D`avs`SgEHx!!2
zP(B_Bn%?r5%Jt5+5PLk5hHEsizt4-o5F$f-@0A{v)N`A%&3Vl)$
z*wG{@x&r7m+iuLApXgxa5d$SCl7G4aiX3G-&FwV*3
zdmqwSQCJkHA(|GPP*aGvFd?i6czfGelxV9EZ~7kxUKg~(`t8t=kSw8WlsNECYD!1=YrAs3h$KVwDI;Sh$kY>zFHZd6
z!JBG$CvEGgl`i6_N;FkCcJ54e-r!C3lZB}NA~P=w-9rIxGp2`>Rkfb#3CJ>=-~R+d
zBlMQad(S_T%!B;EW21nPt31)IM=XD!j|jF%UqUuZfh!!3COX<~jqq*&1(c2m#;d2v
zD(YPmHF$&M9H1#!mb!xR<}oNGPYonj+Vcw_5(RIB=LwFvc%VGB@7PgbM%roP;GpIz
z+=E%VzovA_&tUR$WNd7a8p>{yl%L)n4FkxM2nq-A`tsjayNb}EF$rFm1+j4V&;=s5t`1Kt_-U_jZ+k439tIYOHHy^enS3J~Q)<`t}kCB)0wR!QnX#UFUh498{Lql~rIdYa4!;RPylQ
zWqgUvh1`fzg&KWKTc$o-X=Cj3#KswF1&V)5P+KP;n8KbnvIXT}91a`|hF(2pE-1
zpfZUBogjSbj_u$TbxxjqbwhOq*ca$uK$96m)pL1%2YpS4jo`|a4M+5>tmN$4k|rkX
zQyOpGzTNcnX?WI74AA-iT07UEsOm6`6X>*dhqz5DI12;;N}CK;Bf1dCX56J>28x?Y
zfbjuA(kXC}FqkM~R1}qBrV!aRMsO`~B`z7*g=uYx1Y_ACNEcYyRHW8{eut$GeQTP|
zr{&Dfa(2)E{NML|p7(iP7yrc0#)3$^hgiiQM_^qu?lyJbno{e&?@etWD2>j=CML>G
zM1K86Na=SpHVAb^+h%mM$s`H-qC3N2E$Q_eUSVsaV-EYMjlsbcjYF6UoaWv?{L%NX
z-FU$KjJ9_@5;1p4`m(&94
z->ajeYooJ^La}X;UV6I3HI=4g-#3qCSD(&+|?aB+7XwX
z%M)n~219pu=yD6f2Z7Pm&~^xl_3iLvNC-jGwnX8TfsPN};Liiua=Ad7%{4?_>gzk~
zHgH8Q@m8Li@zYjaCpkgDr2IYQ=;tHFNuCaQLA8~M{*TY=!s7ZV7GBI!>cEJw^BO1&
zXnn}K52^#`=*C#aEsiHhC~ZU~Wpqz6^#{~Q=5ADd34n}0C#g>S-PT^D_1vk!BsNn<
zE4?vNv>`AsVQ>l#gT;vDLi*+=hH=dz*Pz7y=ChceU?hnI7K=rt6J4tYh8I%j_~|2Z
z>NzmA0o}HA|ADUS+bM@C76PfbZ%Qq?_Or_(dik~v%FH5$KS!4I9OL!cP_2#McsI0yhN7PEGyEN!;2
zOPlpQlFo^ne?W-5q%33EAaS_tp>%r8N2uqXMYh1|+cametDmj7HmteBb%WwE{rdJV
z@3@l~`yM|ww+dkkVPB)V~vJ_Ll;8rGEEBc
zZik-BT|%(7iANqw1`jq(dh7>S0CcyQ0s;!)2EzdoLMY+DLTLQ*b?Jo0|ppR4VNR0zBFv^9dJM37Z$>EVq=4g6YLsfy|zE
zOfXf*PX^t%vauq?ywKkMW6iA)QtbPwKaHMohC?tW#FZ4_MK7;$XiJ<#IK_pL%5#w0
ztfc73+gcCXe4ZXLM&E$4WobN4ER|x4YHIO1zu2tUWmXE1Z~l*Vf~6gp)BmFP;J?1P
z@^8f2CV`ogJW!{mU$v30WC5s$m#gMgx%4q(;BW=)l=TQ)3Vi&fbzN*~5zb39tYKSt{hsv!5?=lb}1KsOMf)

diff --git a/docs/articles/discrete_files/figure-html/unnamed-chunk-9-1.png b/docs/articles/discrete_files/figure-html/unnamed-chunk-9-1.png
index 4aa6d9b526975992fc135ee310356e25445bc13a..51a799de825468de3d4cd766c3d03a2a70492dcc 100644
GIT binary patch
literal 43505
zcmd4(bx>SO7e0(mT!9$j;1DbXOK^t>fx+D+3GVJ1gWxb&@ZiA{+$98e4-5S@AgB_J8vl^Fl^>o+f++OA$VQb%j*t!ZP{Aw=cQJ
z*rzDt3FO18+AZkou{%3)U0qhPJEBGIE(^r*Z~uc9fk0fpQayG7{^a8*&-w2|Sc(b!
z0YUw;|KEofzN!Bn`gB|D68zP5;()90hdb;K@Zb+r29HVL529`VcLoX?TtkG>)=zEj
zyV3C_%4S78|A`;k+VoX3sC72npm^*N4NZ5?hp_Eq53l;ZP(WN<3!j4UQPF_a+1*m_628f(m0EP`rH0a9B-(ZE27
zd%4r*NP*6Ht(JiE%ARSWaDE9VawlQuAz$hI$q-(%Y+gOJXGA)A-QR2LH?YcchePkM
zQ9s0+6$w2z?l1J@*SNAR*MEPkRQC??loEEoqS^?CQuEOx
zjMOY!TT4q#Q^^eKh*TIWdZNXxFF?OwXMaKTC>4r1QFSTVk?~T^+VL{CeJ&B)8o$fd
zY1TZd`UCgXtS7W)FU9iNzBdthEr^D{JGb)>iDw#TqI>d8fi(uG`2D5vYBl#mGo05t
zCjSiUczWbYNs(Q$S~APOezZuP-0k?sYuzgCPm*Z3=aVAwW_J2l-U3{{~
z0p`v&IlsE^#y=V7iIzmaBmX?=$=I!jr&zJu{{D-uZ#3f$
z)(6ke_-&|WaRN>T7Hi{l^|<6v8xqWAT+-$B%3Y7=s0X+h*E=na*PIA3^-OQqQdi|3
z_s0G(>{tpBP|lCbQBw7ZWc+@;m1CF@J7a8er?s^AZzWUpLS6CxbaIv0DKXgDiRPfvt
zb0+!qzHK5)ip>W5jT=$7VsfJIC>-zu&t+g+yYT6j7X<)ns?`MEl;d44_p*n~@^*GD
zJ=%6t1u{{4S~jkqCCm1&1ldQcZ6$2_2M{WKzL}%0;O{Rft$e&Q<08_2NyPF+uLS5GGKSnC!S
zQyC~KDh><|&d#zq#Pb`l0i6CnkgN@m#Fh#v;B0na(VdiUT^C-1VIS9gPTI?PD`PK9
zhR8#KgwmRpJ`*@wuAQNHh9nThb;m&EfO&D-fceHE2p??I3^5{@tuFF6r!wP<5w?IW;-|;j82Xk%a=+C5%zCyRe;o
zcg;*?<|VO9Z&bATfhm3DB}F3lIu}XHyD@Vb`!kNNe51|sa8A@ruM9l3PgY;8vd8Aj
z6fTeNV8n0Dbr`;LL-1YUAo+EX1ISB^H_{YzRqEaAkNZ#48XvSzSqhFA+h9rq{|frC
z+L`sI5-Sn18oORG;?xd!i9Pf7I4#v5pDT(tZo%RPHq)8j`JXl7&BxE1qPH3|ZRw=q
z9&D3!nIM27shQj_(bY}&;Ho`f@Se4g@#p12bq$&V)iS65R^e$c{~He~=tgL3=mEtPcwyn@0?ZSqQb%m{Q;rMO*OfG>j9ps
zSpOTL2YAu1;7e@d=IDnjBwUgVi}`M>TQLFDbWf&!wpJ0j=4R1&)1M20AH|k%vS>Q2
z6laIust?+tO?IbM>6dcEacT{I9FF5OBgQVJJ&~un{}Es)82ta+*I$_9yn-nULxl(M
z1Y&A(h&G4fIQ#@FZ}GYbF0Tw~+1kdEe@vy`6Qp3#U5n>cDK_F$l1p@_{lhC=&NFH6
z9&wfBomN>_WXt>13qVD_KScW>^NhUge}jZCckX%BFZ7OnSJ13z_&%GAP1UWH5lP}(
z=~Qbw*0$bXO36Fn67LW5hbFL)jwSgjJQ%(reAs5%%&gSSmS8DFz-@7=Zjko+=38pf
z|6z>^$wtQ8G_Os13q*yw-H~jQImta_TmdmZga)J3Zh*N8JDv-LzAm#8@Eav$JntW1&m(Kh$0oUzfwotO`MrN^%Jix+w*-uhpo+R@RO
z{&^ZqmVJgLOxb%F$0_>JYKYRf_q+Ua&y(+lQhbRJI@A=8Qo
zP9ay-j<+k2QGxU8#_t>5N$bg+!(}evlQAFwXFCMs_8&HLb&pobc9!*^t?P%f(*H-0
zYX>y>?e_?fJbPhP{w`_&nY`Dpbz|K7ASm8?EYU7=fUiw>WICU{$Nj8v
zB5%V(_g*18>Engm4QH|VhpUQNhFZq{_lR8S1FklJ`2KZK?4ybTr#gYyM>_b@^qHLZ
zTlv&TTR6X`W~F0Jn(%5CPvLN5v9WS)W}WBoxA(5Zy})7DXz5IQ$To?T$feFPyksip
zS8{gko6G#=MlA-0&aHVwGX94cr|~cS_Au^lIM0mC6%sn|bdIY*VUG6@>8*BKfAIGyfWUQpGc3x}yC+*!X4JFd+
z`?3x$TQ$Fm#f1<5f4P%3Am6ozoObC01IqG#$^BszBoEUDiTy0OhjX)V=W6AXMrK`s
zX(ny5ArxE--6F@0M-`5p7O$r#XsT5_#7>(qe(@OWi_&{^XAkgvfYLu*3AhT-=?6u?
zRVn!5`%MHl6mPv!+&j0MUEw5>>nyN3Ym#9^R)$t7FVOYUD94zYbxDc!hq0_rt|oGx
z%`qtLTIyI1>S3JM6#X}Ea_?-Sam+gfs0ltOZ29%KRRf9XUxzO*;nw;
zZ_1h@XXN1FN15B#sW@QPsNU*}XNJ-C_dhwckJ_`Q%x2)uotQltqAR2od-Yw0)N(xi
znZ9yW$=KWQh{(sb-AuSjr$d7A(pAwZ+&Z&jw|tAG(yh`wBP&h>^&+PLw9psROA7m&
zg>u3QRNvzF%)M(zUJb6F;5uj1(uT7mPWBY~Q!`s$g{alI%<#CcbSIrpm0Nr3qoSD^
zF$?&`92c%XgUSw{weH2iFLz0J<&j8Y!ltx1rvta6Kk7o~6(b$uX_TX`Q26kmm@VVk-bo#jpm1a6?ZK!~KI_I*KS=ivM^yJx=c6
z8tleB4n*5gE5lMc)lb+5Y8()V(<|_#M>j9;HLGtP*3utHc;%OPiMUUntPW>|@;gOF
z#MmQzwtiXD18qLtpDka$H7;XTdKNV@?q!r!ubrAUWp6etiSOq>hCm$gz>9mk*iaT}
z$Rd)xJ&VQ-*Kr648CnInmDbO~@bncA;d3huD$uUe_zKKdM>k3Kk`Hw7+8Hfw+YRsF
z6W>v(F7t0hASzzLqs_dV8udO}5z*!{9E=rmmr`cpOt;3g;7LngK{Octj0eIZ5U7&5
z<+0JBJjK5_EnmHxqkQbE=u`o
zm&QBN)YH0!9f2s~FoFv8uWx2vc=o%UV_P-n`)?qZ4;~*@({%2e
znLDu}5CsomjYmt$swOW638rC4(ume>N=SwzMCU
zfnY^>AEg+;Q|yy*TxL_cqn0{bwVGFn46na>H74$9U_RQj2E_jKAFLqu@X_2D-GRyB
zV83$wq}-Fn=*GvIU3Dv@{m?*Ph@s9@tl^nksb#KZeFa1IS!^erL$6R-wo;|`MZ_Bw
zKum#6g_`=bth4P5E4{}K5;yqK><-JRbj%$H#E+r`;k3r5w~J{c7JomzO(Xq?TKPYv
z+;{5ve$Ps2i@mNV3ufb!TNE2*hdlHsP~Db9rGI~{e`Hf?O(`9{pIuZvEx!{uEZIZe
z@swCKzT3#@a>GrFw>xU%%&j5hoikH_-`~-U=Mg}c_ns}{yHnQgSLt5IG_i+;g$Wu3
zZANQ#$zGd&0lHIADa7bw!InjY2U=D$TjX1AwG^S^iUtM-uU>r`v%SM^RgRvw6MOwS
zj{Gw9tWz~!(~mKWO1pV>fm<)=Sy;Lkcbi*UpnDO~(a}jste`4unXxgL&r0I$$JADL
zM*pSA#}75rv@!ms$u;%c|7whscA_X%7z%~j*AM8LJ;cf-3$wI?Da6EL$;DA%L6beh
z^9{ot%Q3Q_daF;~^iSLHf+iXrSje&iJ7}UgRnv`GaQhBRdI%>^Z5m7G|6bg^!lTWq
z7JW9G-kWvd5~s@#tG=>WH(CW!c6^8Bcev?~*Uy9Za*of=BO@d8$dnAYlY5DnN!FYM
zL_S?$YF~!DuwfWot=1j4T)nBRb*o;!lOvGb5UUVaOMZQAAms%I!5(!B?N*QFE`rF&NEN==3>&TH$m&A>ddM5
z5^#-+tchTxihJ$tD}B#7jF#Lile5L2n0dCmXVuwjy!MX#0zy<4WGMu~X3XnMMJCP!
z>m-z3NWbzksFUsy4VOm+zZ0hb2y~;r1EA0A)5-DnLyqcj$ogwNyre8QVwRWTk4_dy
zH38y(g-6LA7#PY5#dz!el#I>8m*L`?72RKn<5wBgH8%sI0m1`GPd8SrCKGSgO+q-j
zL2Rb2vC5AB<1x6gI1SGds8y$}l4H_MGTb+6u0oe)_$6p4E7@#L0-C1>#*2!uX(49w
z;enKwE2#>BD!D!(5{luEG=F{&0|9zO3X6TD#>i+krdVvuB$?E^_ev|clUF`JE>lKd
z0St5Q5A}klPw^>~KVH?WR&6dZ&ZIc*OugOY%lPQm5J2cpjIfe-|I}zWu9_{%PUaFS
zS$5z2&yS9gU6Io3dn+Xmq?wsOE(qk#t@h0(?L(b5l!<~hQ?IdW`aTld+aE2tjX-Gs
z2MetyotT5&jKO|44jK;~x-WZ7o2|)}uagVvK6soEctLq7FNae<-*h?18=73_v*8gI6
zAv^H3=eG@3S2{-O=Yv`Hm6s8Sn9tBOa}L+=zk7LFp0smlPL7*Pw_tKqxGdvt5^-}t
zxP1p&a9cmOg?z2s%{2@*m}DxGlHIrVMRgS&6d0M-`KJ^ToU;~O2w$kElIbM`J)
zRRKb`Nd%K*gwsGUSpm1ysr*O;R2^5oAGa$=?TJ#8qtZWJX3;C?N$d`}3er8?Y_cP3
z>XTy2*1FdF*xE1L?UDX+YmQ2(Q~pL(=Nd4Cfi*0+-NR%#IpORpl$$z@sX(m;7a>u@
zJ|^&I7FX@959a}7#v%VuFblxABX^bRBbH-g2vN4K}j
zkf3!->h0I(7t69CU>2o^+ojQ^Z&cj|^mRnGydR$Sc@O4?jB%*cR8(dI=;nq!d8EXp
z>*ud-^D-JdzRucV^QcIE?Sp9nHYc9^Cc-D-1C(4E6c-MTau$|C8l&H8VbJ!L8d9BL
zb*k{{Mz&o;xeh)1XkRR&yd+H24bbObL%(#ziKPQZey3c>XVtHyT*`rVv$AWQ&xj
zK3flWU_SS)5$RD(;w`qnfCzgD(KsKAuBSs;<~uYr<}v-M$o9&>IjreVg&*I>DuI-h
z5akkenm^3d!VX6vb=`<$gQ37aqoPeCAeDi)*ya-|Xi>dE9;0
zA84HG;R??)2+y;|Af(eBPkbo7G@#BwdrW?}XN*I+GzXdn0Xvfw0JxIujks-XaZmd8
z9|C0e8J1G~x&V_wfZho(cx{4k{#86>+XM676&+;Jm<469{9vZhB498uH54tIamsbF
z)G2qFqrz+ZvwyO>GBGwJ53q<07UN)g`?~_oycXZ;OQEQrx#9%-tHx%J5r{2kh?G&0
zZn+JYWS3Zs(U#h@u_5!d%$MY$_50M@$3cW_`q9M;rRG^p_TK6_ufruH(y!uix|LXP
z7Xvr-sfXF$+wP}B`SD(f70aZ<-E`%&!{1XFmFDIQfk(}Jnd$?R~ur2*#Uw4ihOkVP>`5Ec>_0GaetcFN2B
z^KWy8IXbFZwAm7OS#oqkm2R)wDl-94nGFCR0wW
z)C^ul%)d#(>7<5IKkFXuyJ{y(q9zNrX92~9jMdn12nw53eSrIn|{ZaSL
zY*%}wmq1mWuRTlF3D>gHdYZ_UF=#;=8Ts*S`XyO4Ykw1
zL;+yF#r8Gv76AQ!v01c^Rw>SWLZ-=$WJ%aTDz1)@@DHtN13O8=1suovWUgkZvH>o%
zc+Vz%fHvF9Gnj%!pimLGb@UM-K48!Gb6+StJ5r~b<*kl|7RBupkgb7dGR1`scN(OMS~KFsWnIVT6hZ}!*IP#Td3S)%Fr{{N+++WEa>Y0xio?uL-
zEj7zHoZ^A}zYQ)vsft}{%|@i(CwnKRv?%OID4v~=2x&U$z6WTmF@d=UW{OS{D)pS9
zWZIXfNDNCC*c?hn+Ev2-&wdvMk_T+H>0>kpp1GQz9VAdsREJKF1A@8WIke$O-xJ1Z
zH|g$bkf(hQnC>e;TXk0VK%v_j?Xprm4}Z$(wnhbnQlit(3V|Ixr;Vf$UCp
zeA^lr9zF{IIBoRXrR`P~ujLMFAOvb`
zO7{Vl4BEI$ftxC?LxwgtyOIh*xQS7Fa}gxGeQN+V8#p0BraQW0(-`HvgygytcaQ`6
zp`htl4B{BU0eb{fiBYhwHz9wWY)~?G4D)HG2;Q1Bzs-1J4KU&%kP;{l;
zM+FIN3<1as;)Hbhlo%4T!h(@;McaAmcUtMq<*;GryozikSHTfugoA
zam0W%dkR{#zK&v~!Y2t_F=*nwh827IKW2+JsnDHbY9Fs)5Is
zvYYntsw{NAk9&ahs_x{o&)KL$!}r(V`+>VIdn!@q;}7dl!m!_z3^=^SWAIV
zZ_CtSKOSK92)tZsaMj^*B|Kbrfr<5C6+f!8;7{ouPikD(S--YGI)t7O1T!bVKfQIl
z@tJQw%lQWrm8pHO72WH41E)tIQH+4yO=j(iA8(s{uMtpi3S0-}^KNU!rj=XTXC+}|
zeWay=)$%nX21o_0Xd*nXdheNEpN+DBT|b1d;PT)2xjo}&Szh!YP}b&oePDs>0BSPe
z&KkQ=piOn>Pn{=ru06te?ax~$kT~*Ms`W!_4?fzyr$yGknsPil(Fpk}0l9&X1y%J!
zyxErEcrNqk-bD_09dr8|b-Bm;%YL+$PP8rg3GOm8>Y#xIdRNroVX?r6`5MJWHtxqe
z^SXtjkgV$MRoHW-<1J?te(swAqFXttJ?4;du&CMA!J?)aUjc%RYb`dnn5`FwEGuq(4
zmi7pSZrS&D^rb&Ku}x~6tQRz{^Ql}_CRL}?@-_zBl3gJ_;3iA^>b`G#Xk0}FFLcc0
zCKXzL9bk4CwW+tJGFrAW*Z?`Tm}}n}BNv+ptYjd<=lS}jgM$Nam&q3P4AkpTwLBO*
zc&%AdQAsILmZ7#H%0-Z(>8S;`ma=l>n8qD;pr*}3j!fOcGLa*T^*uehb?1j{+xu-3
zwquSLBnn4K6W&k~6PqqZC#b&t&$o57<_@}jf%
z&`KXKQn%9MK0TovvyoFOQw$1qkC`MjHPytOrD^lfU67(>2$b=9KTK@+6Ho74&HK~>
z@~g=&`+9EOeQ6u)VV3VhYp<9}SCB{!prTk*w%kEeaoCweoJnPXGD=JN=v(}lKlmOl
zTSP<3)(M&hi)J0ish8!G*m+nf4PkGng(4!%J|o@aeA0Sow+9uu(iUaie0v^j$vy^m
z{?C0p(;uhVFL8KQSPbPxo-zjVu!f6J{%K1Qc3XavHJ?wmBgH?IC%(ae{tDSd{XLZhWjqygq)
zCp{DU0F0N9;3+g}bn$Av5AHL*q`hJg`j!1T*y99@9HD*r9RO`{CfWFdZ$S)
zt@8}2z1+zkSSKBj%s+})wD#YqvH
zjRfRrua_3$yMsX%We$akB=v~5-$(WRFqko@yDo<37!c6=xnf<83i>ob1c(A0K!pz^Rm(aizr
z?)oyqgP)$U)(j7u%N7sW<6om(y%l$!g=P@c(O9K*-F+DF_26?tmo5-XBY=n!i1mAA
z_~f{ft;H-d-Dh$+j!*3Kt$$17g%#S^_;F>*q)AY3aoX>o(RwM#y$jNb9X&KN6?zxD
z?_z(rmDc?ZM9MeaOG4N@$kw9xaQu|<*nAuJ9{icWEwvL}(RhCBGG+m%Gd`F)jG+fI
zBt2)nA!0E_I3M>+!`AlSZ6b5XRe}LIE~h;6a{6Yy^|Do*a%DsdHi#WB)a?z^IreZ%
zUEU67h!cOTqxj9x+;PnW4luZs3)89XW82z`i>51)LX}&HVOwdsWx>k9E-MIcz)9`u
zu>&tpBvEYq<>m_yngQGZgZM!0fHq%n4#_w_T6TC9b$cvfRy@~21NIXMNF@dWx`l4s
zm|a`DTYjggy|rSqf!|PIA&8+iwU5O@r7mW1q_nLt4K=_tW#~TZSPZe41Xv?ZfZ1A+
zVX8rbm+Vz~`nNM1utLa`HVnQ2A+q(cN&PdGMGp$jEE=bun
z)R4wL70SK7LLzOZBIzwDSu6bP-C8T0@hh%Ftnb&*$v%JISt-9JRXb@clUvV&X3*mW
zwwQjTo(8h9&79Nf6W9y
z1Z!LS9z1l9-%HQXZWFhBd`U@U4b5L~YdE=0D?$)`N7w6A;k|KkZh;DF;N5OZPi*>^
z>#&`*pNyCQ?+&7+Tb@193)v%xGWfF=K3Ff`j)MKO;x^QBrwTef{3(JFp6CHJL?KR8
zXfF)P4hU8P2nT2W&c00w(c=moL?RiTF`I7L_5bdI4S1l=x3jOs_BWgYy`m#gH
zh){#Of`cx#Cpin#Czx22V8|cAPNStb^M8v{ffv=yV5#bWIeTX#uwFY;_WS0-0+_GD
z4PKG{ar(<(`f`U322<-MhFP}~*!4uOL$3Gc!%M7(!g?HoNqxHKjdH?7_TTWVDX>qg
z-nzUevrn`ksYksW$x+ug8bL^G3dONcAJpol%9Sw}IqXl|lwdcIA{QQa)n|k){O=5}
z#>(heHN}5Zzyt#eEoZl!tYF^hBM+7L&;ha~TTaUawN|<_-MP9QC0%w%7ouO_4DIMu
zFe3Eqi~%X0IQmfUxRMP2-@tIYR>J`4*Sm?{&ZbRHU`)$;74^gdN;5uvkQ%9&ERsv>
zR=%<}t~+*5J?THOvfl+-G=QCsJ#n#((LK-z@K#gH7fk~62mdBDg1Ti?5yUMm0{q~N
z;Um+SRo?;gmUbf!0kGpF`tP>=*|?R&yhs*u=kb%{2*%L|fgJFP;Z7md{q^{|%!wnS
z4`@9eQr(~HEU*RIEMQoAIMsi@C%XGlB&+G$epv#Ul{fH$rNtrOFw*euRr)+|C^0@=
za~!SGq#g@QH+LESVSt`)f>-AP8?Ebud5ULs^8SxuJs5zUaZo~Twy|L)>o9Z`wMQw=
z8*7m!Run)D%usto^b7HMLmd{qaDyynPd`c1aX}Ey#Xfoe?&e$U)W?UiML2Xt*Kei%
zlqE9f06!&zN*N#dli5z0a!)?0irS1;KQU`2i+|%n4PntX7un1!nk7N1Iq51qC57B2
zv2f@V!5g*I#W^iL>`8!K+D2C`ucjEa0#Tk`fTAkp=fZl_;n_}7y3bk`hBC3bjSB{v
ze|M>E4AAYVz2&xTY^L3{o}JFsqp%w;KyEf|M(oQyfmK$(u_C?as^&wku)$FAue{XG
zuOv5~D970&Da4v(NmsvwJcZ2SV+zGZ#S^WV^<3dHdmDEpZ23Kk99TsNd&tRb<^{7`
z!u`=<6?lBo@k9@c)wM`D*Cs<64U*(WCwGB$8}Va
z|DHz4X(|qug-RQWMw>BuN4@>F5NEn81_VLmn{AZVu%p;O(bzCQwl~?HLo?1h7gJI?
zWB)mB6qZ!{ZCWrPU|NP)Hq=2={x8K!r
zFoU6j>rxxr_dZ6GB5UV+S5Ol@g3S*-pTu{=8FvP1ANvH74m#WW<)3{(Cmp`#l_cch
zApkz~2SR}9M_%WRTFaA801Zo4H%5-D%m0Z3A5@-xA)7ohUWLnlclwx-vCPDu4K#Ux
z{zcWHAfJ9A?6QUqX~jf82oti~;cagHx28P4Zu~dwpb`tAO1$6nH0WP}mml!9G-)?v
zIl+IUr5G(R7voUTUHo}v^qEfEkD
z6!hiGQvv)JE$dFz+65ArSoJUHb?)0uM#b|TQJrhd-~?R>C)j`_C3WABZR*>)wus}g
zOrR|kVz{>=y%`FjXLrWuAjn{KReqC
zZqo7IAenGF^rks4dt&qooJ>6SXvB-@Y!;E#^qtz2Ln1*H2<@o`Uag+&VN-38RKI->
zbgI!c?T=61`F-GLUX~cEp4D6Vhb$H!A&b{<9dj0L_z@g0O7X7;Fl`NA9y@lt0bD!V
zW&SdvLGQM^S!YD`$<^{plp*rlfB&JRE)-O}=KDN#E0pDd{iaCq4O3L^u0D(qZAQ@D
zwrQu;QNL>#u#vKCQ1IxAQOG~&K7jGe7ssVFNoV>4fJ?tpD&raehku00QJ4y
zWzwW9?dG9f9JpKmYJ;#enIyrn12$0fc0{wFzbmBiOnK}cZO?kUe*>+Bq=Okh$~Y7m
zyj7<^CA>XXD0j77DT&gcex4r6Mm8683!PU-@{5goz0QvpJENG&?dB!HlD+>O^f4MQ
z>9061Wq`U_G=`h3TXxovJT|J}H@RwsUK>g2>FF$5WzzB7EOd19TQg0;BwQ!^qk4ga
zOb^FE@JToyd!L_*hmf@g?Y^}Hdhki;Tpdnuquk%`jc^zzRC_by!bb)Xm+3qb
zE;GUNqgmVp?qs!)$KsF8WR)X0S-e@Z+tZiwvRIgO%NpX(VfMA7o>-;olgs771j5Sx
z_aNg<-=m|YmHzsax}w%?-^{)V;SCw2Eu~IpCU@I5SKCgC-$t$J4{JW8*9ODKT4fgg
z;3t-2Ma|&Y!z{uWR&*+O-J+nMsho65ia(Z#+U=J+Wcul%=GImzyXwg-x#YUY-Exbu
zogpQ;C>AYlwF2!*`$bl8Ioy~mx0-a@-~($e9<%s(6%XaGx6}`|pjmB9Tj~?*Eyjwb
z4?riX7u)beCFhSt+0K0DHIdW3K{-oqt>ys2kG&jk8hq^LJDPw0e(+fS8|W1!dmd;|
z-QnHbt(e?s{CI83aa0F%nmQ8&4rO>Ihw`<`r@V1zigna&w6I>65+9mZCweHMsN-jg
z34%c19x|3IIC2hF{3!dC|Gl3-e-a6Kw}nv?GO3aYdmN4AYbhxzPFNjAvuFvQZZ-Y>
z{hRuw?FVvUNFYMf)7wo;W5tX0
zLc8^en>M__$7oH@PmdcD7H`VhgR>*Re+8X!pg~5;ZQ-)imvWcpH8{7ww$z{IeST~V
z%u1K9S(<~4pkGX?(Qgfj<1*ivs@GE-%#;y1SxN)n@2?CqadUIuxxTlKK1_7RR4zGR
zjRl8+5n;t^=&ZSrTFF&sWNRTyviHy`>`m+mg04G$lwKx-ncbw;HPr1hHhY^>xL*WF
zS=o)@+?e!>S8nF1|9toR%iZ|*69nZ*o-1X3GBi|MQK*M
zts)NmX_LDNMcngJ)1U@JL;cE7cEoxiu)2LKxicJr2!p9QPwVRSvRfkUpKjboMMTiy
zk@0SY2yE~k_K`bJ+)W6w`TN_St?uYi$qc2-^n0KWi8wy{g-R@s)kFnYEizv-(U@+b
zdb6(j`Nyl?=&^F^DWDYzpz+Wi!S=t+fpq)8LbkIli%Hs54$Hvz)DmYp*Jm&
z-?1Z#DJI>pGx{@kO6UAd)vJ15#~C2S6^rmDHqP6D*6mty=*mRpu88w%aq)8VQQ3jR
zDR0UN4&$DL=@yP&!QI~7O|b0WzhBvI%0=*YgZ825!?6klI$hVAFnlDxW|`Z$S~6Vu
z5cO9%uC5H{;w+6ZHLgr5IpBpIp7HgVJ+{`>(ZU$zGK;bA-@kiP1P2G#ZnuyEcU;u-
zKHFcpgp-M&muU&5^j`b(Gm1%V1Z)OZaHw+*73c`))p_7NBx2bOd$7ttHc&aD?2fi(
zK3us4fHf2q@PJVfxYllxUbbQe9x3*1G9?P)V!|DM--s{5}|SMdeJQGzqcFGKL&A!@ID_)d>XkwJU1_c@=7risvA
z-%D^L3jijF(d!TK0B)emEaT7Tt8O5h1Bpa7&lNKyukbjn{ZaVwrtzg8K61R=I=S1Z
zTGR*lr;eD1c1mB(vEGk2&lh_VvHPpeM!f_gr}A|k#~^Adz#~ylRI0EW)c_JmP`$th
z3IE#}I!JO)?mT+)=jj|
zaNh96BTYJ_)-61&!73>!eI#v+*Un5&*Ov4Fhq6SeY>pdkRbI$*3LZ|_JFbqL%|*zS
zSxu5mL?k6)$}G{B!0UK?#vt~AOBncEKxC@#mu}X{X)TU0BE>hoXse`1A@~h*+*#;m
zG6eyLbNU6;!I~jolwQy8d5QB
z?SynK>)PRp|L8}JYje*+3N=fgxL7O3bUGRZbWT*-f6Qp|xo`uh-+D21ww=zj?k{hm}P&&0Jr&%{unObTHFCCid0Wit*I9AdR5z
z-MhzaInL>@)HmHj%wZ()M7v_E;r%747}m*pZyd0e`oPy``UN|KAQ4simS~O&9Xld@
zEuQIlr0(VmfU_~5hx0U)2DWEgd$A#F5pu#ebc1YgXNSfu#-YK%p5WL7Z8sPIcqi&6
z;S-1Vy{Pl!l_N}CZ;s*@FA{hyi1aXvIWMyiA#-rW{h;hi3u`Z{7{uP-@|6o}`srnqy5DjOVP^e;ZL%u`(jE`m=f>;Y1q
zU;gFySsj>a**^=%k*qrWw!HvO47G5Bs_|ohafsq#k1rua$19v2r5-H7hd!vs{
zv^6l83nJz)8_A31FqTDUo%kVLJ;3gokf0zaRxFr!GoM<&K*hw266tx^g0!0{6aVmu
z@)B^KKq9s}hh#h;9prBwlS50n$n2?aJeQ&#j%#;6ObqNO+PDE`Q47KyyID_3PA&w}D9Dfu!gC7?+_ryzhmZ@gwwJ1qjIbX|vxnesD^{NBFdI;v!as4jeh#pDbghf3m7PA%>dzGt;?@ZH$LAYI@sG|mj5OSSec^m5lt(4
zLnuXDr-v#V5Kso$ge5Raf2x>}kPwhUx7er~`1Iy@Y3J0G$k;Ss8i^oS`o29k?1uY#Pl&B_qhFM4VSqspe+0bJ)wnCL$RW6ghH-IsgNoi!tCXN^C
ziJ-rxiH8DZ96H?mo(A#`d=}tje-&7|jOF~j!8?6@fS62-=8!N(T>SWC3k%)xy1KeG
zSv_;$Q&~K%PzF|{QK2VkxL?tlTSlF3S<^peCt)JnulGtYD1ie5izk_G^-#034>(n3
zBqt|VS6BDjPXA|Ae0;Eu2E))&Q&S5HYNN8UvO+^c`)#)zJWrOh%J^q{MR3wp4SJ~{
z&tC<0uQiGcewNKg0&D^mRUp{3EA6-}$4kJ)KXjopX0w?{?=~JD
zsD(1l76C39`Vr99@Mh$PZb*XTTN-tk6d0}|s_Yj-0s}jbcNZ-W)AtSz@bU4v`5z2?
zxD;>|>AEww42m>Ad~($K;qGp=$8ph1*$apUn&uWW=QLZPe0RFZ{f-FhZZ&wUUCj~`
z1rQJS>dBtI@mcS1<`r_!k`)YUe&;GFeibqOwr@A
zDeJM*&XD#xI3OUP3YP*r4^Sp3d?Fa+7FaV8K3TMnUMy;ORw;lg%zQ#;qA
z47e+g3u6oOG=f0^ga|{9X>lg-JDLHDp#UC~d14Phn6T>B@Gkp+umq*M3^;bj5lSgy
zwYUtLj8iS2dB#us8fkOWS8H1;>XpZqpw4|Pl&6NWm4IO7<%cc7q=0#N;(S0s+MU2x
z3@V3*fgpStVMsA3(2J?m3iYUZw5&L@a-4||>s>U{$h!ZB`=J#Qz$1%7>hGqOuhN+y
zZlw6_=Ds9mfg&!J5O0AyM1zsBg-Z()yt|h?`~`GA-duMN^aZL$<^aam(R7oDm3bK;
zDm+zsv<&a)#v5<6({J(Q%JbEfd>$wqd_e?&*i8TO)NGuTGt?W^R7ZkQqW
za8Mf%nU7x}9_WW<`%!K3qx
zL7s=8#RmEU0(s@~WeKyfqIr;2fgQf?DHM;@I4lonmYOv;H@mvJ?r%(@p;4%oe>Z;Z
zNG?B=Pe`{ftX-cf`w7p#Wqt4vPgqzO0NM8bK39A0=Iz@h0#6ZC#eu=Y{c6{u%P3+TCd<|Pqwd2l;)5$HHwjY)Wd7qdf{|Rc
z-UGwGVt6131N{BNLHTx0^-My$>?YvV+qMHtr46Sv$Zvp*Adh86J&Og
z@0tV7wwegQ%NknsUS|N+3Uq5p2npvG7u7)C0A&idHD;ozsfilU76Ush_k>AF>PMt1
zl88TqS#k01p&IDkijvq%_rFoflm^vD2k({-&>10dc6`2@>WCqdXjg@KXV@KzeT1kT
z{#WqEx9GTf|5p4rcP3h~YJ<+*rgxRqQ1l=W?NIf`?n_xOJo-1wt8AULNo4=A(!d{R
z1Iy(bLoB~Swjv$61>&v%#FRZZL9)hOHogp2##5+Ow70aBdmZU5jCJ#WrV|7OlFv1F
zRO(yY`B@oW@tB~;(U!6&8)!UC->}HW+S*LWB4;)#1FaGb9mv28f{_3(-RzZ~AsUp*
zOeY$7LPE+c#XZkUz0YfWTY$zyC6m%kK=8=%*p>fkSn2-r+F06IlKIaTev2^fP|R&d<>?aw8HGvO;N-ctOFhrjGQ#bHW;2yMosf&`fYLIA*KB;cr8H7?ufp}^+~
zf&ZhtuqubM~5Phv&KEY&PLg_fMy?}VYay3
zM7`FnG<{Up-EtruNkMSzs1(Dh14t=@DQLk#7%;te@7{w0w@|^6T-137R&lHl`_CrX
zZ>;;b0O^thk!;w}r0cn_f&_o@q-56tfygQ^I39JhDnvan?1;7=6s)X*6GqTn3!YA8
z>Zi+T_Pn1dgxmo987F&kclVfZ$IH+w9juK72ft+iU^HF0K~f4(V>=ZpUq$z1WUy1)EXvOlsg&#ETar1@c
z1c@0qwlJ_c=n{qh`TwIVaJ;bllJX3F@5hzg
z2AdE=wbfoHjKL^%=IHs(7yx3`0N4O6VpRSvFE8H*tQ`?@hC2knb{l#@U8#pAgxL8h9WomI+^
zi~?o=*|FoG4?J~tdO9&VNr2l_p#-g0FahA@<({s97=Zo}xu8qm!8|~GX?iFGKp!O;
zym$uK4*+by(E1O^dXq&c$;c9bYlEy9#ir*4C~vl>C&q1+g@=a+31WplqP?|s@%=N3
z<26(rFX;XS1_hA`y4Zw26051Hk&Jo*^2aw&3yAFLI9vvP0E(jOiVD!1$_5Jn{lp6I
z1^}xhJXRLD)EF^y;yVQ&aC;h831R{zXj!ICEx)d_~STD+3gp)3}}d0G?lYt6Suau0DV!*Q8wLO=mx}jW3+H8
z0;iuC{p0IeeTb`%}xgMptMo}Xa>z0?5x6;SRg((CGn64WLtz}3f4tpib1-o
z^a9SbItu=nRdCNm@haQdtDD&om7qfiz61+5b+}`E-;3tlIaE@${Rz37FlTVD7x=YS
zxmCtCa$6a|Q11bOIcU1{YfL}AapOi!mMp!nFy^4%47_;QuQ8ClyO^)%xxa+lStvcb
zf6|AOre94*h6?v22{~neLa`*BS)<4UXsS#IKr(R*DSy|nn|ec#^G^6-=c0?1>#
z2viN|5X&;4=*Z!xXq8JK--G5qcDt4G{=It^*xm%b*^Lb=_E=Cak&=^JmTH2m`|`yL
zTEZvFkGZ)=%t|$xqe1UY;^cI<7ZAEJi?ZQsOlk$vf@$kVAYO|+D;<^<(XHrSVJx$)
z^k23iV4JTMyB`UJ!-hbN4b=&k_2SOY&yjj)gJ%?rV{~3>0KC8qe`jYc1vV!tBcQ%N
za9Ec=TBah`z#rtubp%B)D76Sy*|oK`G47i}#baw>_Iz{!SB=H`r#&8%B%w;@G395E
z9h!xnZkTC$FJzh62tUA+1%FHC7&OoZoTr3RpWCc?U%L7y$_cbOQT$G8UoGm_?|lSm
zY`m_7>pD0|EAxNR_SR8Vw%gn2vhW2&UV{)2P)fQ}%Ay6Or6r}2Zm>WlB?TmuMK?&N
zK?_I=i|z*L=FDfU_xFwQoiWZB=Ztd>f9&nvYpv(GW8O2aIp=k6<3At#Y999Tybo(+m9x#qO}1c{<{v(LmfmY|1o^A*l|b+uyzB+y+GKZJ=6>H8VEPFdPK-h-s3s;w
zW7or{0H1iQl+T7;#FRUvp%0jzY_1GzD^FyffU_J#&JAh3FCgz=1H9Wm2}{5Xq{X`d
z1_RsklhfLgFA9Q*dD~Iz+u`Dis@3Am9xyXRFB+GcGyyDZ(eY7hd&KGHV@b)nrVn(G
zf@7Wstpvge!AArk$~R)T>R&YjJ0vy|*|bk2h+iZ}Y{=#pNArgt*XEvvGtSSnNe5mN
zP^hU1Ay>)hn#7ZT;^}z^xe|aOzCzISFiv0?!s26Y$UP{OG-5EfX=%s0V8J+aL5nE#
z<^$fyaQzA;CG6A8%#1qd4C7I~RE2A@$d8A3UDw87Pfx(}3?t>#I}|FF+w}S^zgHJe
zyeX8YM4-I2L{Gw@Bc#^o@{d0Wk*u4Chc+82ba>X$T~8OAQ`<0?_3=1bd_y!D7!1gj
zi1L@$eiz&}^RqPgoh;8xv2
zl^DSuA%i9)l(+hq+Cj7MKwPifzTU6^5_XNfz5D|^c;NU~*p1(vpmuTD0W4#Cb#%0~
zmG#!GTWTGG#ndNTvUf5Q;?ZC}2Cj?tQuUpC{pJm1LlMyV)TvW&%HMUk9q#@GM8&S>
z36$ea`G1q>k>FKPemSH=*5yATHX!P$ukdqCO_%+BjnI=2bX
zKJNf!3Sl1;vxvz4FzBJp4Nv6^cQ578Of$2qjaP{hJ=bNy8i0YRgx7&K0jQHvoDX7@
z*#r-pI^R4&`1idC3e*Lc75*^+cCM<{sWLHnYPMo+|CO$B
z<)uCd#J|EQqXij^Dz^!n=NE(c^q$Q86*X
ztGB=|8&r1QQ--4!t@`EuyPskJx72`Y2NW2PbTH6E#)MQtk23WVuc8+QuGOD<@S_pb
z!Aj+-)Po1IEGnsi%c&Ia4i69MAI6ImYvk!|ClaE@sIt}Lb1})j(=X8yUaVDRhUtiJ
z+*D`L-?v(u<>le#apDmY7YCh9?ePnv5O8HrOREx%q(4r^aTuYrb2=Tc_04vX!(q{e$xNC
zJKAv#`~gs`R*BW15E}rOHEOsZ3+KV><*<#i-6)n_6mY+
z@Y)K%T?_*!U>{5;)5R4?$$?X%g^i>M_xJXKsKvm71YN&fVn2NnBe(bQ#|rR~`~`pD
z30o9+RJ;58QP7h%I_d+VH|Q{gqb-W<3H=siV&Mz{@ay)sp3Hi53(z!ocXw|<36zZD
zzcqS#>FDY@K!g$+iX^1)h$uw=;^qvIAcGU6Uf;OUGKOb>kLLdSEG0xRMvD1*7Crp_
zTd8Oxx5b)W{ws?BG~>aJ=v6vfgN^_b=B^T8P(Fy7%x#$
zT2>~@VsY9w3r8xUUu|)M)m2rlO7YF`FK}5vd|~XJnW4Gk?_~AU$3p%$%6I(
zEJdpE4h2LN;5>B0&=0%hIR_*z%!gNN@clXd67u$~jN8X$0m^0@k<+LqD@AXgl7_)=
zCCf!;YGsJoAqYcJwT^uGCSYU;IZ9??ctIi>%+>%BeKwF#TEDaD{ieM`f23D7Ko~xV
zSV#cu%xpVIHxPr#6HzF=(%>FP8ReT3uBo4T9y1iwDm4aR^Q+0mq+bf8l
zio9Z6AO8Sk*NXifB#|KbA`P&0OkaY?4{P=x4|Z;LoUFa;d=B$lyIL+hbqn!kq3#Hz
zwgl8R90OhOvshSKIy&+oF9LX!^+;8@zrX)IKxUfkwv{lt2Rj*IZ1&~1kK*b?58I)t
zyDmmJ^;tfIDF~EAs(vP&Wx7Qm`gnf0xw1;-poY(qi^
zapMmsmB~sIfGXDXq{L9n<
zz-1UY9kWt`C_DuUI*cKp{F0muPAHN!0uu!43bvBpyvrEuGY7}TbOCTPN9QZVGN0ti
zh%Bg7?=ReuLh4&3DGExT?ZSLuTL2;VyKrq{dOC8m69xlu7YjZ86l`dv^D+|O!=si(
z=>T#%C|^9JcnYMpEODy~Zn@OCIE0R(U}yg&IOVbn!<9QwXc1m~K>$2Utx_9ZNl9Py
zIYQDu$obfJF*&K{aX7$~5SPShrR(7vAvMml?0{QC4UIxq(j8{qD^N&u47%y(Hx!D=
zlwUhn;GeD1ktj#n*)0AUCoH^->WZj-a8FfL72MoJx0#p-e-J-eB+;@2*;Cs1gYxOk
zSzDWE$LDt1D62Y*giWe
z22uyF1shMlcW|H&=H-VU5jYg_eCE)5@3!QrjPmUfIa+?ff@zd5Y}xE+5{?X*<-!8o
zflqBaAC*`=ku@!P(uJI;<-x7Gr;vlbbLS4RXOl&-_%}C+gQX%UNr7Aha6dMVj-Eio
zeT7kBccVkX49dI!S!EKUMUB0yfAF)+nZQY-3Qd!^sAgt-1`Bi6aa}B{o-QzI1mt`!
zlrIFy1VBBD2JHGMM3p{1KC)nlsrfA;{QVmtP==^ICh92)b$+2rn{B2HU9^(IVjX9I
zTd!s~Kad%n45i8_M6&MS8tkmC$qYeIsQvlw8`u$~-~}qjOb63W`~I5$X@<$9-n|R%
z_V=_l&RD@LEI4XF4=NcTklXaqnc%R?x&t;8zRanSI|N!-#mVh~r)L#ixEM2L=!z^@
z-}w74lh+bN7$8A}vAE(j9em=e$oPc8I!h#5F`A5bg4L)%Gs5Qhf$zOPBP!4ukKe%f
z;>^H^`iP2I*hL>aX9ovB^`+8KJfa=}c=v`XoaDQX1V83KOfCaGIZ4BE^X5b|={r|B
zL+oFm;wk9~F(*0otSuO#RGJLgaprKP#dnFYV2>!y4eO8#M}@|7Ypx#;G(Jm7HF_*W
zKn&mAtp6S2yOos{+jw#uY0-dl@y9s$qm;jF?35jqt^Ps-+fXKpV`ui+5>W9AKOfk<$^fiz^x8+Z2dIzvDP{BW{cv6hKBV7
zkt+f9m&wSOnjtES*O9JEDG_7%Rol(9o_txgBH1arqi#>XGUMEe3k1s~noI$Th{{UwRigf;{YkMAT^@9V=?R~iWQyM%g$k5-&0CVY2
zOp^fBcVP_jil2o3wy@^O?60?S0w%f|`WuqH(T)(f8XqWq61u@WU;>^@V};Y0&bvIk
zff6757`;O88;IKXohp+cybxLOTzyy1n#777}~JI#3zw;;rt(KFw?OGGjE
z;!5x8u!>#&_NvmWtDvnJ!{-r|=CYElUt3dKiv&2@U!N6#y*oJ1shOVy-a6>nb^wE6
zkKlY8b##V~pmU!IIm`s2A%qw?`3(>cr(V^q`4_xo1Xpj-HVYJdzww@wT?F!NkRm9n
ztULh>9b9iXV6g($S`d9*cOpPZDo4%z`AB->$yK^}e>0K(oxI7ZX7|z^#iiQIUP!93W`0{ON|9Gr+
zGeo9hOlx<~S_KOcG1|?CIj?$e=9fFraE~=Y5qf{c>i3N)xu_nnVKycHa{!t{4H87c
z&kok%AF!N&^%Gvan5&%f9t{zCN1R{>6eJaOh!YYKZEbEI9_@^d0{jR%R{UFMsCY{5
zvs=pOrgvAEICRSFPVPw@uDzny(9pn$F^0*GEB}KyP2SB3?{ABLF#gb5{$aF-ZN(BB
z(!Ut>R%wm8C=^`(r9GZT%&HFcFu`gzBLN&D08{9_pMbCeuD-L3aryO;8>*FIrNDg*
z+iYe^gtDwQffFf25)yhI#@~AUenIAf@}!~=>KNYepY3)3>xTJGds-AF@A49jZ_^YqOLLB|9IgWR*G
zP-)^gZg8SP$h(vMJ!zcoOtf-J*yQ`Of}30@2ACQkYY;j?#M=dxYz*^&1h;__D|`x|
zqJFu(NoFu9#6~^_qC^?26o5Gk*$fF44IDyUPL5{5cLCNJP&AZKEp9C-BO_C23&{eh
zL~$=G44hNv^LQw9MtFCL>Y-OK+O>Q%M!fcZ&6ETjcHguVM{|Z4p|wYsF2!=wrJOQp
zU1;UJPQqdWHLn#uW^fodbt|rtk~(iKsL8?*cS4#IC44MGq&h=wQfwCPgNF~Lr;PKk
zB;el@UAVvoQz*9Rg(#mP1O5XU-cT7
z%Dr=4p-^srJf)LlH1`=ILUxJYUV+Qs<=dIued`@klYihcX!D5t-+v!LEv+oltl$%v
z^Gc?-b1CqB8k5t5SLn`U>79k%){^Tm-1RBhq5-xH$rq;q1yG>c2Dv(1)IVAHQ;@ve
z@dzeYqmV^Fgc`fWWKkP_d%O@mTAkTDvPm9o13k@OHOlg9j?@zph`k+w7gDA&PMJp7
zo_-(!%G^e)+@R`?my0VOa*ttPV*n(Gv-$(?R`K9#O!0e_Qt{00`SZP{q7HCvLG>v^
zSq@le$N>gGU4yv$=G+-p2EYYiuc73b4w6;D|6D-H&^3-|tB(J&mwnu_P+^x4U5a(;
z>hAGbXL=p<=
zfNUeUfFKz+i=`br-GyWp`ku)^?i0+JWhz$^U;=bffw$DNH6g(WP~?HJ*(nq`!-TQs
zH`x=tr)JXO%|xRzHCL}dr;mP1WcSV}tEdi>63dM{58gMu9o-wPa00Y=4j80i#{=KL
zhxA<+7#VtisMN!njUlI_DkIaBbmtYg@_kUx0x&aT{o4dbt4>CuLPROLa+ugCs51aS
zAg(%d4n=z=U#p{cdA(}71g-D2*snh@JO}4>?Ctln_1Z&FqED%1nDaCPO6+Dg&X8?FfJ`sFl3Jq)5g0?2$!?-{!llhft4k
zeQs@FBU(_l!f3st*fh7&zZuEq(;ChFDVRi}WqwQ30!{nb_F+I^kc{|@N&JNFU4|Lp{wdzPqA*A94D0p<~ZfYuYNt|
zXI9Wc8d@M~|D_rYkq8uiatyRXz_Z;92EY&b$%#pjQrVGvdNjXWSIFK={di0Zw#J~z
z0BO@YDX!gzLkDinR>1*uX1uYIX-|bOrTmQdvGio
zSI2V!Y0T!SG;-QM`}2r>zD#;Ax-^gr3mfquG+|9Wih=qnRbWbd(#3K;4+^nJk?$4gTJ`<4=onw>5R)e-5|JZywB;29sS*^*
z4BG|d|qrU4=jP_XH))MI8&B8@RV;5i{Xn8C47xymn;{7Sv!BaZXNH)Yhv>SP-
zLZd~Aj%810BgGAKrQH5Sjs>g&LJ1KgV|8>c{iw4&b5U0CVL5^Ie#;vl0CiAfLWhp{
zsF>D)Ehd)S>*ko_?Fsmknvm@
z(6#L;`f$LY>rBK}>hvCqIgPzWJ}MGg?;VNWQH%hS+AB+V7+@@jjcGq2t#8HzA2e3?
zwl6w?VOt^a(wo!AU+}$1`DhBh9ga*X3q4_WD0Jdhj}nRGQgZB(+J1SatfwWXRfRuRm;z2huf8Y%X4iXA@m`i9&XMguKOH)h7(~K
z>!nh?#f!s%7!lI9_UY~Abiu*$7B=(8?KL^I))tPm@LXdK0pY@MTu>chdgO+Joh5WJvv2U|q*&)YyfU*PBh=%Ms{W
z)ghyqO|!C22m(+SkilNl`@Up**vB4rm9@k6WmN4zXCqE6R3U=$>R?03A+M&SEu_$A
zeI;zY0b9_;i1fc-dt!y`ndsz-+@OKQX3Kk*!=s>Jc&_iSAVfCvKf2)UA3Di&Jvetf
z;^Q4(!H5MlXp0yMk5y=+WOhtD
zoOMg?D+d-{_(!K^SKbykA7bo$YT|{St_*jS2sfFukK1V&^kY+Poy$vgclJAjH!};i9TZz(1e+J!0-}18B^mu?`44rL~?!^|7
z=6xDrm#6xA4={N(ZQ~0TxRrQ}{cYD9Lj3rO0Rd?Xt43i?5*A}CTKMf)0rK1Z@K6Kq
ztYYn-?6S8vTiV)RX&3Wgw-!(tnhXL`H|=qy0NG*!cRk=B2t0ol{kJD
zFC-hLmbzl&kj6o@_Lm#?(?5u$ij_OJ4OIj+TYTVq{3MC!(3GSB
z#SgXnEu$HZQsuNQgNOO*aPwtB>_Nrfzm2Im5;7HZwWURz+B-StcZxugCH;Q%;G<+{
zwwH^7sZ20SpRdu?GKWdj7GSr432E-1mvk*lF|ky+TZPW+8LpMK>rR!DJ`ICT_(z2a
zMSFd&^T+h6Ls;-)xo}F_A>Y6pawzMunZ-{Ih3BXrfeOvn^vInYeKFN=>JaDJO4xyiwVV!I!
zw;E+();T^p9hDQeYQu94dt~kSQWa5>vAbIut#bt>ZMnJnN?MPF%oL5V=UoZ8C_vO}
zd>r4l#F@KMe*f;ODn8`0FC$Rw6e*L6ECeRLZ$fX(YROO6(c<4nYfOj-c)
z+t{Y2-5GY>$uA7-8m)JeZ};U`;?57V5zJVWJu0Dk=tP8hKJa~89=FHxfz)}dK#9YI
zT}7feU7m0$DOU}R{?$L+;?dX8_)lnPK;Fifg&Y_^8);x>JW@jrICK7&tnwKaYUN0o
zadkmTk#~<3RrhydvWWlLgwvESI*4HZ^JvS{Rj6DAZ6_N0U#|zV?*+~xrFdu(P$Yf;
z^-gyc6awjqv46>v*j(fA@UWXwysWV?y4Qj~Mgih3k3Z!7cUCKCglU--&WUI7+3ua^
zu(262|9B~b6!M$`aS$C?IQ}c70NyTCj?r^+a%!Zq3qy9{G6_jNfPj$UadiBOlz%~S
z7&*|%?{=RPx^K!!0o4vc!L+ha&{y*1#?c`Iq5s>!z$q~&9!N%oL7C#c(Y_$c9DUCl
z;6>_*ZkNnva`F>=x(mKw#Sbj<62n0sq>XpRIEpH
zjVU0^NK{O0Y`IPz;n`v0zuzhbEQ)Ug;CFrdleuX2-K)jp9C$t+w6XR6u^
z!<~^yP^!MVUmBD)M$4s?(S>ccUk{hPhY+q&R18AxA}ZXvDm^51q1f8zs*usILLZ#_
z$VG*GmH{0Ky_WM&-+lP{Q|#+-l=DC#?)Y$FfBHE;Rm@>nlwqEeq(fw<$Sjc`>mP+)
zgqPNZmkGP+y;vp$XK&-+_jKFf9xl}@LxyPM-d}0$Zbb7&m;^P$uG{kgDQ>rMXb~5+&2tQ)Wx>0C4
z#9R_+nA)(ET1KJImLGMQ$@WSKWBoD|niD=0(2mxep$&fJur3Q$Bl~`
zj6_|sW{w7h3$yFf5VUs7^RXY153;~*#*SMN`JC?lCQYO*#!&XJz5ee0ghmAkAJ!8M
z^+rsL+yoi>M$GnSYY(~34(;}AW32N=8xJ0?um8cGnTr1-j5VWp+qO3*vBJ%j9OrqA
zA_Ke8o)Bc+^SzCV)jV>(ZGEwU^p*iuIB18FU!N~`VlA<`*YS2SY&&bBK4hjpvrYda
zEN5ME`88;!>`L6-Ei+{o>qrc9)O%ybrzJ64oh@aC=oB72M2WU{6-77ZT8FK)r1=yS
zZ$wY=#{{+`iqQC99sCQZ6U!l?XyGFInCZbbk7o2k6EBE~upd()8}c(#t#6#pD(*-p
zwN!tk%=I}=$6iA&BRy<^Noe`Q_)N~4
zP~#8sl-*&}qpG>8%{ZSx3XCD`6--&D$eb=_hSqC+C_nYvw+1}ylSPpyKRG(C+{%e!
z40zI|<0L+ibh6KlGlo2fmX7I~ZtY{ti@UihJNQK-Eh@I$BSWVg>vJ0jBA+WcevRel
z)vcnPM<;mQJ}6{g96me@R0PgCl+Zck`OTVa1o)jbUST%3G#=BI8$#``NRZ$o5oz#;SX=^8%j#aDu_>_=>O%E
zB%|G=pPrBI_;{G}{Is{8zV)8s-Uekn60F)u0qbJjK>g3Tt$rXh32V=si%V7Gt=Z0o
zMa7Og9F}~nCtKZV{!2CVwkZeGS}$#Tuk_^v;pWW=BUiMZl3~|lW7b=-HLtJL;}kZ<
zsb5rD12{IH;9`XRIXa9fSzGzOPW&oq^BZpxc(-dToD!joL?3nki{M6hrH@nE7X#ek
z>G|K1-iAlF&n-PEe)jo~rY!5$%r2WziD#N&J??Yk(>nJq)Xfi7r1%jz9aW%@GXH$z
z^Xa3Cj@IK5L0%9l40_3q`Cqot;^)9&J7p-`|tvkkl2wz_34xh;_IDM
zDkLY?MpF7TBdcy6kofJC&>Ohtje;fML6%!wd{5?B)eKgD94sO?CE+acWwldPYn70Vvs=sPY&OxCGt8Svt2zFN_O~J0w((%TT
zT>e5QO|PqboT|lZQ?ZkkM}gherk}91#>QR;7Dw)Y8V)LqI*~tKTQ2=i0oWiylg+HM
zSx*KPDfpdXhB>T3*NkvytlNlN26e654=TR?ln6fxx9zv8uu4#1G(wOi6e7i!u8SC3
z540Ey$j2sg#3e459^7_=OeZ2wa5I2!;Ox2yzqQ5tDkk$HJ<1Q34Y|cF3B%ZNqv(M_
z(f0zQh!))V^8K7FlWN7isFah^_8iyexT)_V$jUYU`Y)mXLYv#XpFO|BlY{BYCz)zE
zd85TgK5wjCU8r28ufNs0sD-h#D41!(&1}pEbaoRhFn32Vhh_6P^$oEcWp2&mW`^1U
zYOc^rRs)91$0gPdW#WMrq&#kgSQif^N)GLJp)2vC=5AS$0m65>Jm1*lh%PnO+ka+6
z&UY7-furW3L~_ROp$^@5hUKo-EAKsv9pVe2O&1Y1U;igX_Qf0=?UIvI_d>vavJYTnV
zF@#M_k2E+;r%}p9aoRc}5fv%~LVsGZ!uB{@iX8;j8`hoO+^Zerw;)G+Kqx6hw&!?h
z&}5}+xMc5x`L1OE=KMLFs;HYplxO=VF-nq6zc^v_E`sVvvxblAIXD$Gh9GP*N;bm;
zd885i>&S81?fBz1ToH#eDzqfX;3@z9)v&CY_F4Xl-e5V@jovyVZHzOvzCFn5>b>Uo
zU|Ri)9H-lr&4a|I)3~-mUsW)9p&UAM$4_4n6)`Y0S<7uj8mUL~b7Ku!lP^4+d^M4o
zmGBpZ3Fi1q&*o>$VrBWYI0${mPB4|Lb+;A{PbTx)Y}}cW9M7WPKStBZad+KGdF%P$
zOM2R+bKY+wiaw1nC%Li^+!XG(GKHRFRzCBZ(ctbC!k1Ui$XqhVfAMg@*u}@^!i+t>
zp^QL`@rLIi|3*wq+wI#i3fqyugpGRtF6oqSbFg{o})2`e@JKDxSP~FTpt7tH>8R
zU<{8^hir~E+Ry!{@oLZ<-bt$!OmvJcXFx`
z6)}=yvS*1wivU#VP=*5^$%5VxW3$3GOP2>i5-thk8^+oQMwAnZR+
zX{AjY*EWP38x@SLD?0v2H{fJ(ci5w0P^0u=+4F<*Xh?6Pd^IMBy-+Dj6ZK~VFTNG8
zk_yv~Di%wt&-NT#iiOWn5u`A*l=*&rrJPPg921p61Gdrq`Dpd({zHt`1@aS0kN
zQ^@L?FJA7ZJj;|I6dh&iY?9J3ZJ(m%w^|f7{>eQBI5lE*Pm2aREJRWPf!{b+C4npf
zG4T$xAXgG3ea~7`iboj`0F%eK*MXLbGJU*Cmc%GdpFZ7k
zF9bHwJn$H}vjntFB$!9&Sy@|IJ2IR$W_)klV*?iW;B-CWoQj>j{mI7LXjG9rtCQ5p
z)qwhm)m2+(_O*{gLpo{Q&B9xZ<>lp4f>sg|5;Qc&ITH*+HZIQ20{YzKY6;}X%>!eUs|$i&k$^GYvX^37uVLn`Lr%BVlnA6w`~E;`4eU%E9hfr{TeY(NdG*92Rfey;(=Tb%A`Ur#+RwJS^%
z0j)hQZ%J#o_@8}0>tKm!mq#`5*i({jV%}^mbR8XuC=8zzltwhGqizr5;%o)pd~bnOD)GFbF>(m*;~(6@7t{L
z21uxNB$$#xdp0Ub2h{Z1L3@JQXpNj=R0J!=DE7#0>Meh3@8?$oZ?NOdtGe;*BNXQh
zFrY@})}F%U#774V10vk|_D8gr94OiB=QTKb4qeX!kI=
zXRp83*r-7hd5XrXj(;UnX6@#_orgP83bsq6f{LqzxA+sEPyS*@;0MV!_kW{QZJaCd
zh?MWXq%UO)=bLs~e7)-$ME>-@u`<5IOS*j9MbYbkg;_RCrLU3CBjMynfOwzpssC2w
zJ8S_<7M*lYvj6sSc*N%$FA*gR=BdBJ`|lwWlYgTm8HyY;DxCS1{oo=R
zz2;-{%$h74A7wI!&P-XqxadCcU@v*G=d-Ac$&Cfp0?qjElvakmb
zA55H{8Kz|EqxdmdyUhLre(LlJ`-Szm!4iI)#*aL6j|r7Gp;y6{rlLV-vMWnOV_h+y
zcv}VmNl_8kU*gqyOnI(-SX?<(BYZ8)87c!^7~4ks^7X`uTCwklLEbTOsGp`y{}g2LW|HJ`?WSEbs76so`XR$
zcUzAF8{Pj#@(y`M@5;lBVCo~A=AoO-%!}u(O^LL3p2@r&FhSNVeP~XfNL2>E!)AT8
z>zm9q9+I+&4;|FQQZ2|k>O40=L_Y?cS6ZoX@C^B}!DcTnxARU&+)}m{94Kd9>Waad@#f4d24?O9?$?cg4p8(`RTAy_{iore1*2eR)W{qc=;YCs952gcfW&Y
zWd{=rZQW1>_9+tXe5mP>fui8D&K2k7tm7!2Y~mFU*FEB8^%@!MXp*WvOP_fZT&@!<
z$m!CfS-W8tgjThmXacfLq7N5!s+dmEdN-;v_
zRMq6jt0MH_Q0zIZeA06V+RLVC9*U@pRc)dX@5zn*&dwQioW}A^I(17)hodbnEiJJx
zhF9v|dy2rm&8GAF9Nro?ZpM#B*oZ$ZG0OQ`
zr@^cuPJNPG5TihCz`YP3=!%`P3T8q_J=_mFVy*D;kDo>P!6Ae~D)6l@f^1%Z8;<`z
zX#2#R*8NN~>vlpmrQ+D{F@hIq(z<-Ss)?VbQX3Tc_VI*L%|j-I
z-+eh)M-|1w1)E_qx0uIoj}8kDDyG_NR~{;IpZ;>!rsfHaXHpb7Rq5V&mX_)&?A6(m}cwL1R@o0UOB}4G)x&sR?LuZtj#zde`j%M5Z
zgbrM{3ip|Mltghd;hq-{&vS)Vc?w}=S=olz?#3N4`Jtan8XPS5yT-o73aj*H#%SF@
zMi~LBzwTaF*JgC>tx?T)*FWnTasK*~JQJ3w(f=!kf~&5F06AAlo7|44Z*d!C-w$VW
zGh*nvXAzv=Ztn1On3}&-b8{|v|F1M|Ya>sKQTT@K%}|24=GE}@>8OvEe|w(XrC*cu
zNOo(X+_e1{Qh?uc8T-^FI7qj0zw!l9=k(=lDElqn&`pl7q?SwczVlf<>v;5}d#24?
zw^F2y9SPS^Etl_{vD+emnP>a7X||brJRc^mxTEZUaj@`#ZPn*G$UfRV3ls}Cn8$yHY`iu)yvPq~7&{95A7~8F>@WHA9O$RM2-eD!t)qU-X@1CA~OWr`GX3@h{dLQ}j
zElxPyqb|o)daid?YqExPlAk7vNX#b`XtUD;gtzLW;rKO!1Uo0;d%sbK!AjztDoWEq
zk-HE087&s1-Rs!p;{@E7`JLtsE!>bNKH~W6DxVbUb>KJ~U$LVHQWv>w9b}P7Gi%{G
z>>efT@YRVGdyY{$Ktk#3x#SEt)WH>rcH~6*3)@WmlODsGIk6+&dLa@V`uYy?Sh=$g
zN;En|ik9B)HdUzWg%0`a`T6Use690=QCjYavD(
zu5Ur1RJqO)rF(u-^x6oze5G4L{=!JjjXz$@hC&$&S2h>=8i;?2J$0M9bnbkdHulV^
z%E0&e6hA$>`||STh>z(#s98)a<-Rn~o?VMsf0!u$?3N;n7ks<~Zi@-}fSAPF{
zae|h;EDZE>m8%4{5}6|JYN~@rwx-5vM}#V#4D@3a7jkh{qFsf
z6l&|0e<=Co!Vg*+)0PO&AtSMs`t3`=!7!~{mT#ZJp~sfbt@1+K(ejHS1eetO#20$<
zpAH^I*i>~!Xd&mO?%)z%(&<7>WCS`zeyg5L#2&r2eJ={na)=vz^q`er%D(7@-mU0H
zxJfQktN2$I@)F;c*GZ>Hq~+d6ar73Nz=e~-RObXt_s`}zIq7D4<1g$LRpxe*U;6yQ
z%W<_DZga1xG10@isjA{{d8UKmBUTTgUf*EFyu_Yvd%9C4(G1KyWs*d!?
ze!a9{_buhonuUSzW5kz08Gzc4piEP;{DB86*L*o%^oc<-rqMlnp{=#hP+W2KKutLM{xp$nk^-!$dBtzZa(wPKm?{OC5Ei#Sj)!c#WLCHya9^ed!w@l
z-s20=?kqefsWWYGz4`vOTAHGXCU%lz@Bg7N@B^VJ)Y0Q3ZXZ#a_P~+lEw<~b7qk0Z-9BY+CgeZ+87uh8m^EYlC>!b~GOSPMIDv*xS?93VvI*
z$%U7t_Vj7=fG|CpT-5f8-96tQ^^Y*YQetlY
z{rYU>kh5EdPn;NfGgI`eduyr(i&Yl;2rkwPV$UKHC*Bjjyw{d`tKZzg((deovLRAg
z#Q^qJ!Jg%)RtA41$tZ^3%Te9Tl??Q)^NaS)t2A+tYxv5#NAl7B2
z=9BI=+ezs@czdgCroqyqzjxA!*&81_U_Lp!s7JB$s)+a;8&w=Di+up|X`h_FLh>k$~t8omsD`FylyGg?Xvu+QL^6PuXTih3}e^
zmJ=>fC1p#RP_*8tn9Gv?AdOZI(5%8V#$F$8=*x@FvF&)9RCuMZ5gG(ED&B{xVt60sKG;!luy6fT5^s4?g-6gBbuds?4<>Pz2jI0eH)BqwP
zf4+@i0I+`{U{SGu)L^o(f1;p1vHx(r0P!sHw?+=(G4eOY4G1OjS7*cq7t=3M{Bnef
zX1Qae?-<4P+TZ_Re?M0D+@tY>$U_k2kBX95K?{sj}{cKO`pJiw0
zs(aJD@bHsjTI=-0C-*CWEV1qG?fvbL%^PL5wbnzS*!qla18q+mhCpteY9B-YfZD*5?`S4DA^j6JOo09pm9uMb;zm
z-hX>~`2r9gB7gr6H~#Wkpi5KI**07jYd62-~UoQ^Avg%ne8Y!hrk+*m8=YIR);R0NQC?
zhc^Bc6~;#cB$LWVz{wz&L5@0?4J=o{xc^)vULX!w`YoU_K=}E@eS3SPbqBiIyr+%I
zvre02&}*;?swaJ+dv=SEBYF{dt+mSRo&sIG+V$P`+pEUL#y~Xe4%Ac7uM3n|3DDy7
z;>>wyrc4BS9_38Z^sn!|GN>iIb-JLN?a8hW8wfOY7IFF}+;RzAWbul#@H&L1C$4SA
zz8uK@NN8wIfIeFQ$n2r(MnmXRCCejIIoD!)qjQKBR2GCjUFap6ote>FegQli60U0o
zs;WId-(3M7GkRzWS?2WFYkhl(4ydo5!#RfaQ%)yKN5e4WM7$5ndM&C~Di*WVvlLXljDSay*|4
zr&5-I#3jrIxW7&>Lw97XJemy=S!g1*jH!&%G((7mom;+qV3PcikMm1T`RSa!uyg+~SwdEWO
zWU(pzUcbICElu{U)^7YI8k))f=UuhPyGSVR+}m@4o@m)OY2O9Bf25|S#y+~@dlwo_
zZkPj0baQjFTM*nM0;H^N6YrQ{ca8PTeM`%ML7q@hgen;6bGI0-*dhs@&LxOQ3De0L
z(mcf}392ey%tKunxCviu=jU@BZJ)n_>b~YfT&|M~eOOT8B+kOJ0L}40uQmv+6u{N4
zR_pDR@cmpw0<_n`)hup6ISCw3mCzTSm(S#)R;Tj-J77IuVTzX90rkpBaO~w-Mk1V0
zqPcDxGtgi15w1dj9uSNE-0S4zf6AUu0xL)^AkG*4CxJ6jpi;8#He>)GA;QJMVFQmx
z_}Zr!p-N7~o$rRMtfuOT*rC
zV$N*>Yt9H{K0w%8YUt}0fdH5Rn)>+p`GtgB;no0PzVDz>q-fM7;d;QGt8SC+OGf+b
zRxLcm9y&rdH#bi&OUc6o+inp^KE8JSHjs|P6+?t4S4brP3Q+~uo*b>hvJ1pn13@(0
z$MFw7K6A?pIQu;6OIcexc5oQ}^;3+{hX(wJ%^$=PSD%d*nT4QFpM;(=c%Pk}jkGQ6
z*9^^h#V2x4A^qrw+-F_ich4h~xNv7JkQ(Sg`#5X`*^vXGi&_#J$pLoq=U@7*vobPD
zbpbgCBxLlGe~kaB@v007Q5tdBMSZ+QOKSj4p3p1|g!^VMKqFa)^+2y7@51S-ZfG?G
zIqU>_Ft|#m2&nrsoY?+Pk2Zw9Dj1mq>^9H_2ISNEKp1$>@4~ylK)yk8qn76SN(%S{eXeol%S#uwyld1KHlv>at3X7Qay~NXkF^k!4%t@JTHvQopTID%LmwlUJ7_S&0Oi8~iS8Rv
zBY1dta{DuJwm|DZusY@_(x0bq%Z4z=7joKzmId}ygstK!C`|br$Xxg$paSX<6+-l|
z>hmy$hXL-Q5fakug03#G5xGDZO?t2eyd-X*Z(x8Ok)8~|4*)H;5fO25l`x9$x0i;>
zdD+?9G>?Hi1lY{AT@%r0dZW*v#2UWCfjG%ezW+GlZzcGik1m6g0DyuFbS%2l*rACN|s}$nkIZZqh375@Bcn#+XGHDJ86P
z{FWMgzz|E2Y~bDrrV?HU)mSlX6HssH{NRD1LU(9POpM^^n@XmpsUO5Q!IFTUhK5~M
zDn2mvo;$-XVc^@qAreb20QaWBmoTeWYhe_2M{*C4L6q<811TxF3fI4eFW!+r8}#l%
zR-QAU7A!0*e0+eK+Pn`4z|{Kq9DBh{25h9Qt(>)N^T3_dKz06{pMe$30j|aX;tcSX
zfJO_B+UjWS3G}6cE-J9x-q=`#hT6Xmh^d4^eW15E4)T!WFR^(vG=|iS|i^
z{ee3N7JMpsHFNsFT1(%*ujBkrGl-`L829Dke&D0)$3ZWVXrI_vTk+@XciJ-%*u+
z^5jXFP0ij+RRi4G0^Y_o<{x9m(7KNf{ggeYn^PInW?6aj+2ZGfs1`7jO%iVgc((wK+v#
z(q~{-07D7PqJlMj&>p(J>Y#RoZ^-B5NUG%#^s_T1dfJ043c^(l^l`kryr2i+&Xt7H
z*~n3%SI=F6pmh|SZ1ANyv`eyK*TO&u5P!AU7D8o7e8`=k#;$?KH40t^+%&PV>T{wU
z)@uP*T$yc@AawHJi~}(z^wEP|<)ZKhz5p_^e(07UK%W3
z1gVEg`@ksB^I-nWU>SjqR0v#j(0O^={Gd_Z2f<(c78EM0JH>c%+6jp+aR*R65431uY_r5nNbgDG(?K
zDkO!1jI9tYC|Dh25f@A>vJD6YtBzs?f|Vi_kfMXgCJ2I70a*>5hAN1HvdAEPuD|=I
ze|mosCimX6eBb+>d(QKo&NS19p{q(htg@GNhho&A7@6O6_pa6c0Z#{q>YQZjY@2UV
zbj9Kx9Bo`c;CUA*jM!ixk>5t?j+X0DF*HvkjT}{%WS?MD#9wcMETPtbJ2+kFQLT3`
zdu^ld1USl3FLf$Yfr?Kb-Fk8@COVp?O6Be&V(UY+OF@hsu~){XB9YXWQ9d(zenyD$
zk;X)=wez62ekGlOqEiywy4g``aJjH>n44fTRF}BONBW$>oe=Ok?9`=7WjOk|DcH2u
zox%TE*y9t7VK!DMqw1C063+-YC&B5e|5{cR=28*Y*Y+Oh{Em=f((>c{9r{Krz1wE(
z7?>@RThcL)iTP@Yu#-+EC58TxmzsQkW_AZ1pShKeL-zWOJpwKD{`md)ivr_2PxmrT
z46Upp5eT(;KWI1DxF2hczy0CDk#ciqSeT6RZZwp@$uCt*yZe~y>^yYnEjL*XQ-18)
z2G!S7dKT0;dnHG_nW3KF3dyuap!-%dPaYm<3*h*zYGPkPO?Z4aGz8CTpDCC;ZHOz#
z763^t&o(d3*rLnu!@Z%tp}~wLo^;**C8?9g)CFc@FGr!_h!LcGesnvdu5<9CGYxQil6So<~~yqZG-^
z{!js2U0ormppnF}6HhVk68!L1m^|w3RUn#Vn3b_Rr1h+w%ygE6Sdy)t*?65UVK0cJ
zr2gWmGZS}wtAmRChWf1%rY$`f0}q6hVVkd^rIoqA(YTacTD-iln6L%!C9`oG6%Z_l
zv1gB-lZ+cX%XtXx#~}l_SWx9WeDDBixO?^-inlxtcE5`ur|Ha{o0^eP!nt!I%X$M%
zJ>%d!;1flG`_l~vDG^ylSurh~l)T4I9PVtcu1mzsJ$}c4BiuvbH`obtOEuD1R#v7m
zI(7-;PFaQYTt!77vkKYoePN%D41_K=Y)e9DPsj6eJ1?*}_V)H~IxF5PK1c!XEF
z1{x{OYghEA_Zu4CvWG<@?0V-AsyKrG#xyQm*@n1JrE1Qh-qdRs)CA>Q^43l`*Ku(o
z7tY44j~ctOdcMU#X{!#=4aDNPOTY%v`-55mU3GQ23@{n`Fl1xw#W)1l=_}E^RvI6E
zM6TVN#=E?8oaVTIxWFAO;GKJDX%rP=FQj=SKUpiFuae~^mB%g%afcM}o7zLCa=)v1
zPE7!9UHK&XW0Du)J+w67K*~`tD2Uq?3(QS$UgQk^&6lB0Eh(j)Lm(iO`Kw2(Yibr~
zXjxD60f!w4%ccq=Gsg~BstvQgW;@B&50T;b=q55#iSOqOo*$BuVxQ&bTH{CuUq`6$
z=vmmhGU%ESM8!Wp2`cVb-Z3;rT`<&ha|+;dYuF|YI85zGGVta>)ZUaHu!o&!DbC?9
z($-!oSf~gL!{Iq@Ybr<`EYxvHYERu5)Qr@+kuF%gNq)7T)#iWDpN1LPo?sSQ)xtwj
z!nwYM3Q5HcF1DT0UrNM<*(oJue4wT#fF$}-8gDJ4ft5OGc50%A%{V=9M1M5bX4+`g
zs@**IgVO#C1LOz3o&Dk{vXx3=H}&BuPO{yux~Ko-tEM@LwBMROxtZG{vZp
zSxOy!e$$28mQr1#3Wcyg5>}$w{p!3
zIpIXewdh@(p9EJb2%$MKaBl>UC3wW7LY5>KS_HqWUzWTFv?lM(-(20?3@rBsQbtd`
z{@YvTHF`OuB}~W7Z+p+cNjy5cr2UDGho|LoIm853xfH09Hou5OoSyhGmPAyFrh%1l
zOx(l;=C#s>%TLF9P^NL=^;N-y2ChCMwo1Ie{^zy4f3I(ZC;5-d$}@|Ho2^ST(|+m@
Pe(d7-rNhv*U`-^|`LS_xbQdMV|B$?Ii>PL5fs(sD?nC
zb3-7`Jil-n{H4eE;YILs(OyBv5rLo&fc`rbGP`DqKrkVY5ASQZ#V?P!x>1-mHcb_C
zQ(j2ECe9&beCo<6%Dsp8>rbnlO=G`s>BTvlg0IiMP3GKD;8J*suR*Fmyh2Hl&dF4B
z;b!=z?F3^S_2s9-tX-=!u15O$#RA2RNY++6vaWkCHx7=VQ6LZ(&R5~#3q-y69a{KG
z=Jyjp_`>%~vcAVnp?Nzf!`(9ad-1ltG>tcaGy*kMSDj+I2xWGJ?TmTEVX=G@b=pC?JtYMa6ZG8r$I#d~a454!^nQW#qLcr>52u&aR%V
zk)>AWv29(T3u$T3Mnf?mYPdn#s(Q_Bkw18yp;T
zlnx(RAu}r?p555H#}L9C&^To;`ncu(ZxeHKM-^^X2?;$+hnZuAenv5qMYUl;*Zg$6
z?uVA?wLLG4K2KDKnE*Xe!_>~fn4QRcuj1)>4a}Gw`6qSZPQ26lx%2wfp}Rg(My{i;
zS|+s3%*|C}1(9lFmCjv9Pg;WMiE;S5rx5izrY0uQ!M;q>@wc>_nwq*L&QSAm1^U-X
zcVv|b57#2CWF`C+!7E7bF|`%iHX732>hp-Q*AUE#P6g^Zj5a2~~F_%+Aiv
z!ongfT&5Jo1B_yR&yWOdcujmJF*i3?ru*`hD_5>vdnr!#hV`${dq*o49UUE?x(Tyl
z$~#Z$ov6+#>*~f%5a#1`k-ECNUo74S2a|oCFtlH*HbiM@Q2{aDV^IBKVK1I=c=>!T
zvwooe^769Cgx5wDq1I76!F9}8d#OKfZ|E!}4BuDJbp+jN#8^dBH@ZgZ~-5!#mUNyXKvA$dwT+cF5*+7O9G9
z@ldoUIsEGF?*wXxf`Nk;MzjL$J-5F!u6GP7H@`kjT^GFeuWe?(i4{FEhIyqkFcALlCi|LU9+q;Q)|7XQCC)NCmZb$dCbs5l~$
z6&Vv#G@_aIUY`dSgJ7C@3c*Z$QBhHeW2Vb$Z!rCEh587VQ_=vBZ)?M({hD6oZxJ!T
z6y}e9E{!rng_d-v`_xH34{1A$d}>#4)g?>cKSfp2xZw?l+D!cMLphqg_owRb!Ep4h
zCRM)lWlO6xTbrNvfy(O3usY1KLT_bzX?|ostCfMKtxM|>L0&5NmS@B__1PVc=h(JY
zDMJ^++B*Y@wxcwD_Kzn~Y>(vU1<9HMmx|go@?x{KwfnL;j0|0PjLsWd!s5KZCgHjm
zz_{7MD5m*+dj$+%J|yvZms`4E&pz$&6ZNGf3uoIS-6t(wJ4u`=jdD(+PnV8s*_A$f
zYM3P{5;Px75^S^So((2nqM-?v^eizCVdggQSl8+9&eqigcBd<(nG;z|cg?Z30>R`2
zi;F3ZULr#?r{sm7ZPUjG(kIKqbtk{7D4$p<=aG|Wf*I=S>ODXoAAI&UZFA3I6VLx@
zT9EMd$HNMJFuH?n%N@Iw@w^((Qpe2g)xN1lb1#pyA`WXkQ`>_B){u^^%$i;mbcI1J
zKNw+D+bqfZCqp$f7p?mr=IS^)n8zv4w&z%BY8E7_OKRj~^rUyK;m#m#2EZ{PxoN4d
z(s}>*AYS^0NZWjvNz^Ts<>~OuYM1l=9Qs!ZlOw&o({q)D>j)FUH>fj}cr}A)v(@E=
ztg~HgJwIH{H2M)5?dH#BW@_2Rurik8ec*6BUdCwppnAkjNQHu+oTuSm*`i2#oG>I`wxAUlX#InWtDMpwS
z8z_|jRM5=NZHwrt7Gjw-A=tmTd6)K<(AKA}_u7ipYE~AF6C$P@9GSkNvENn#+75i~
zNIJWHU&>zQ(ynaVN~aYp60u676^focC8LlEg;U>CLqob1YMk10BzhNqnc6=9jU1yq
z{QRHhSbLy@92ldR3!D}n2RMpqB)_rh6&>Qz$YBk*O)ho0sU=5+zAwwsUzwfVoPns&
zIVYG-Y`P&O+pHYAEO3(3J?=i|NE#$_iiAE7cm`y|;?ctgVi(E+o3l
zvFi@t&$rrf=KWj;E>|a3+NdRXHka#5ugCBcy}I)m+Fxz;^hVKbo0nqD#h+9-Wb3Gx
zTCwvz{qQ0C>1_%E8x3Q@u3KNOyT5WvaDvpeABmE~rAnSVi<1ok3|wkD>7@lF+kLHu
zc`H3G_`rB&sj3b8phw*XBTX)rsaIJX@TsSK9ZdlH$PkbL=-v{-7)5dI+O@{CMk-8v
zJRLoKhgd0QX2zs$0)-t3q+ov`PW*^wZ3vko!A
zXOP3S2E}HU{aoHhr86$teRHw8=oFo=Y0@_P10v~bYsWu+zKS-14Gx^T9cas>C%mWR5U+WLXmmWT7CulY!)?ZoOx
zvBSXUEzjzz?Z}^_I+}4xi}hI&9y5o9#$jhfN;oG@Iyqx0E_R5~0~F8s(*se+v+CQg
z!-bOaC-|*p=N7kS)kQXOu`kRMwWcW2mF9m9M7Q8L-_ZxiUcA`(?j1*v#w>bC7}Z~m
z@;b=OyHosPy+fWgqQ?iR;+`*A`^1sTV}7)*O%uJOX-mAd>egt`L`-ovnnA{=;k|!o
z{rmGdFCM<#x|`B~x!!Fo`N`t$pF!89Fd`D}BCF^6bFYh+b<1ozpZIOW7{0zjI{&BM
zkU(JcnvLc!9B#(1jn{2T41IiNQ|E3Ff$(3{*jVY?%uiEU^Xch7cNqeYRQ3i
zDM%w9!aweo1Vitw{;t_(5=%FGe}mdOLMRPAA?NdT$1V1b@8`U$o(je@hM3Q@DTVml
z&HAp2BX%cCs$^06Ox$ZDH{{a2qAO%;@Xg2jWurVKvH#aT>gRDi>HF(_gd1G+4ms%6
z4_S1@6cycu8^*0a7ILoTU~@D#{U1hf-v~#4PMJDgF_tAj#YoGS_2VNA-w?VY(<~!p
zE@!X$^a_7mnU#n9vMqLYizJSM)K~MV;JdRQ&x2mSF?>9)1poN!mzsx2?JLv;?`Dj=
zPUaSx|FN&3*HCbc@we}(AU*2VpeU)2{(-Yw9@J=ei^0Ki)8dq(;KWK6^{0W(I6?Ul
z=^LeX9_FvhY$HNLsXzR;NmVo~^&RdhD)w|&Ox&9~P-+UL(iGfVtu-}VBHW1#oRFs=
zy&o2l?nfMgz9;6;@az4rmzj30Pd2X}O;G5K9FdNUj`MBjS7}tc@tg-q0I>3#%ufWL
zQOdpD0zX&bpR)twHMAlsgt`JHEU(YguKt0xYjOOcw*Tr)
zkxT4ivu-9aZ(dW!o}~@Q+AgbFEVb;rS@yC;-;(aXP2&F4meu$@78c6}FSH{e-hAs%
z-~QSUE!A#JCmX}lI-jg@j1q2G4eLYJ-s}$M0kFFw-u~G(7Aq7DhwO>v&a2a8q8rf<0-}cs2J@(RKr(N%r79nH*
zn3NhRaC_JzPPnQ#VJW|g$Lk64e;d13dYZ+qeGqMWjVERDGFj~3PyYHD6RN9Ei!;uX
zfK0(bT-h(p;_hup1Et8_=1&0cU0nOK^iBsgRu7gH+QyDKW0x}GIuCN*rkwi5@zFn2
zrW+DczihWL68^DYC8O`rL*d638HMx=
z4lT!=oTF@n9r;v&^HQ?2$4ZWD?3o;-@GcwfWa(ivof7Si
z@@+;OYGPfN|KvWI;*wAKFWnqgc^_=|P#P?P8UnN*WiXQ!CuFI(E$C`9LmYiR(XB6_HN=Z|#SjA=>QM^O&bM
zHTfDEyw&~lH7#vHc$nuqlcvicyQw5&?2i}InP#hsJjez$$&zc~XM>zFz2rH(|l{nFs0U(9i3>_<;KZO$-=jHOW|S>!1ZpNog=2J*0rd-KkHnvY&4
za(?vxR^o~E{N9_&sKQg}?x&32t9co~;idh@Wrsd5-+ei|C*1%{BVtgQ2n0ZxP}|Y3
zoe9oErizMca$S++&J$Dy`=R5lmNfAy&MVnLH-tQ0%1cNtXq&cIG7yYxa3WPb@)UoiD!UUbtzpq)@tB{CnO}e-+RZh9udK9Ey%iFvp%a-H*8d}
zy>ndVyu^8}#O>9dXFgxTMVU}I1$ZYCB7a+cv0^?PgHh)jO=o8-2nY%4tJ$%dlwo1M
z9(3{0FSU&&Li)Pf%+I5Yw}=Z;P)g)`x6CPPe`lc5+3E|W>S6>{tY^`SJI|hsRV}r$
zX=aCae?s^k`aw*di3_9%nTb8+udLEgX4V!;9by%EPi}7EQ(vEHTG2ot+=JoAewmmo
zB)AHiyw6uSGW(Edos!OmrR2;YV(h&ga)+l=Uj`vl`v|HVT~;yc(mDw^{4g@l)g9`w
zKQ-fjm1T1wb#XwIk8iH$6ha1midERQ=k}3zzMmu8(w?}kw}b57y2s|zt;O^O3rA3-
z`gtFIY*9^Z^U$&O%)6+pFJg)&K<9zq_N}yB_s5|=PJFVyoH;OjvAA8
zB(~Kkk6auYS)wfJ5nBs4*--NcM6#bYD-g@&t!eDgMwQA!$(CQ|u*ZvZtVhbr-`8Mh
zYp_8e4zIz|3oS+|4fg~y3Lmd+omg~zh=|eT4e?mBY%y?cH4}&?MfjrS;o&_v%(9$t
zrZY6{{HAjJmQ9NxfBf6eZ{{hnLIT%G5WbEN;iovZy|$_{XX#0MgyWwTS3PLNG@hrj
z9BVV?Kp;8;$DDxF5DGni0?a6_)Ue-Abp3_W;~3-DY=>4R$}MORi06J+pnTq+evojo
z-b_7=@)&2WOQYXbypeR2J2bhU;dlyRe-mH{0`WLjYP&qg97{HYJ54@7Rgt_K6&Bma
z^IVt`fuR1R#0DPuCF=0-%<5r{)xbXY$pF7ib-s4qnLwp(egxt>4d7-71j(~+GD*&f
zJ1P#D-^FjSk1#~+&wX7acjZeUM)(HGL30Yt45r8VvK{(`86G@T9hwbu;>y>5?T7o=
zn`12w+PGhz4pl$g&DUP^sXn%i?cLTSy;MR%qitlgfYo%-J&*9!gbR-|exFE4Y14>M
zL;UXAN5ZkR&+vpH&i#=lj)pW61Y&6tqP@QQScmjVXZ6{RjXLd?`ZSW-_jf#d4$ATk
zl;zJLn0{V|hZHz!N6L`9M7u{Ayyq0GdtUcEddOiLRZGQv22ua<-!pps`I=EY!$qvl
zZz>bY^&bovq2lhvw6h=(T1`;(#dmV!U~AYKq^eogWnP)qs%0cK9+rGzKsdRdU@hEt
z+5B4RjQd1mmA{vitzNCh%+8^BTRvM)q;U$Ettnh9Xee>X8!qXNoH#nOHQesOVNb3U
z!JT7?bK2_5z@0*bF2TxrGlx@kJ~=8%PghafsS;!{UvVC=TsMi}vJ?a&bbluXDdSE3
zboZD0i=#T}xc@Vt)~i5*GRF<@kONCSvBRY}*NLncDzA)SHa64id=JK|;#7zbGKnu9
zf+AME+}~<_t^vYJ$(gKceaD)l
z4nH}O0|uFzIT{6tcwYqK`ag|Q{~6?SCgg*ud?-^HNGL54HkU0h^w=?|(0>GGW=sMl
z;q~qZv(%-Qsd;+NfwjB(uJiglS7@@X-0n>czyoWK{s9}uy!oMX
z9n#yZ=LJjiQ~ShjM2DY3_^R9ELH1ieWewijKL<&ScFnL5g
z3`T!Zr+2>D3~LaPeT~9UWm&$W4^QJaEwQ12)NW79lK|APD{pDJIQX>Kj9oE?ap{ET(&tV|SW<&{`Pv{?=~jqODtY5d8xAHM4bQ+{^`1{Q(A@~7DG
z?3EGcw~@+arJSoL>X=)%x5hmycMNd=^)46)gJf9zEix-BEBPb%IMErqZ{?<#%=ne%b8~^T+Pg0k;Fv*)hM=NB!R^Y|M%X9I%n6CLM-<0RgB-(^=
zzP4Y74{NHd9vi1)eU3`_ULLY2edMK-Jw1_qc(x34*^8X+QXMUVDs&>yA1o8Z1T!K3
zrzk)}L-U_fu$Y(_2ggTH4Q_311=V0XY)ey9Oy`>~NF}BA&Q8!HR1JRg$RAYYpAEQ=
zyRENaw-!=4rbmR7^OmeMJBJNx=o$m4UknC5H7pQeVVOPM$^Ia+Wm?Q(<~?&%yXH6J
z`?yz#v&ufakZ+x?Hr!gt*_jyF+V?0Lj{JSXYU;LQFy@*BS?N1cxVV^}ku6Kqk&t7k
z=%MXx;Gm@T;7x{#=FvXn$Cp(gKbG;qqLOsWdi|65<#SX;yfMXP_s=NDr1YdKvjTJ?
zf#SHYm+hE%sm0BS&M~8$t-gIebK3$#S}Fa2!1HhZ
zYuc~W+PP6EhCC1d%0bOv+_}Vx65ubQ0ftv&@AjyifxIedC0lh>>6zThc?OWsELfyg
z01n3V2y}!9)GK<(J6?U5(_1OV{T7#{ZPYhd2>I&FdH6nOoQUeiszU5iXH0m5QEs^{
zKjfvMfsiJAO*@i4CpfqgTx%GK7(96E+W4$1+oD9`z_Oh6G!#jUl>?=aJY%)Ufs;~_
zM6v|>32#{-U>Q5OT1U0H@;k}SD`lPbDZS)Qd~4
zVAq>nM;_kM>mk30jTkB#)G5rxKLK;&g&(*j<>BPL;i~t8L%F9}O+-Ui?7I$Ao6{Z8
zJw2o$<8vsf>ZLhsEOqyDWL024PwnAtJ*PCd1(4(e8>-$r@+jMBh(w8SXBuEI)4bg4
z(;3W?uMgA&k0UOZtJF5fI#aS+mQiVagg&^dWhqh*
zUN*c6neqi*Jl?8LvDB<)GkQFnIj&>-SfSL`TAUgT0w#UmA6aF`3aQV^rTQ<##B1`_
zTIFLMH6sNK-0d(fA@MzbB^OT8n2EqWN{(ESW@NN!BQTOb39Cn-wDH#H>4K$A3e;R#Qb5u4?<6x5&tBGSpa;QHFq{=cu
z!cEk{sF1uc9U}Di^YTz-POz4eT6cL2*sSuVX>&kj7i%3efaT0Yes};Pu;mZ-
z=UMeU6e8zJ_(G=Q&Yb0F4&(wE)8qq~M~1l&v~U?5#XBuDmABP&Fug_Tk3u37T-kxe
zvlxB{1LY$kU!04ZJ)GVVv(kCe!h#>R7JB|BUIhmbvHsj2O*%kKUTuDj6~e6rRnO)dQ~Agc)KK*;)M*ppNfB$h^ai_MqvD|PJ(M6~{A
zF8~Yj)vH($)d-H5VBm{yU}u@hl4t1tWjWr}w`ay3o)&$*BPO+-R|JyJSt1zsB~aDL
zQ-s~H-l{xSB}{udyL#`}&yakWEXr#Dz5RJRjLmvdhP1E#U^AI{XoXfnt0OKBvVk`r
z-~su~w)dq6CjKS$eo!!=bEpI~L+m%5#oLlIvqQtf-C1Z*g{iu8E>LC3w
zX~R<Mt;6!wW3ahs=+`6;Di|Z9vjQQURWI`aGQDL!w%2lMJ(Y6ME4;;qSp2;(3`FJZw
zGexySu=Nl9W%94zdsh5tbx%uKY0;OtFFW0RgW|?NzyK8lFxFkEBaPdJq9u=&|D(Cu
z-3M!Huo>&apzy{9ueP=FS!w|73F+-Biy$gdF-jPc^US%77XVuzW-ngSnHV}&`C3as
zO+Owqp_vUk)A|J%wqnPs`WCeap7IhCDZI?=;<}D=3m-Y*nYzQiaJHp8)#_Ew34Y9b
z*s6%RLO+&YE{HY;lGQo|+{Bw@8^E4}t&>2>gt^rWN5r_$ucr)csBjO0w4u*cXS6e(
zL#yOZ-upXVv(Z1xT01ovSRM^D&B+&}*O^9kU!ii)ozo3(xBfOhx&~f)3EM&WozzEo
ztcaPiCVAW;X0!f`$Ep|(dX*1;^?OHR9+KQkNP8t=U7=^p)q2++z6&aZpyAVb`dQQ?
z$CbZb&BOUwvMkku7`P><%}ygi3)X=BBJ|%U00_w8E-tp?$-Rv^9Oxa`#DDw*1Icx`
zI?$=78|yJ%Ilmzb>&PhR;Yc;)V335PVHI?SOcVqVkUrK&9h9X8m`pFzgFk*8YCj-7
zLq#Uwtb6Wjt(&VjR8BG}W^|1uK5#C_$r2i-%vZRYGyi<}-Li-5h~V!_$kk}k0nE8<
zwI(QvUVZ?-U$Ji)(KWG`Xr{~5y7I||J35cEK#(4Agm+Ysi`%Pgui#>DFj`J0Ft=K}
zie;+yI}gN=0rfl}gVZBh_3N?s(;YlR7C~XXeU9d{iS^KTxcQH8^E$gnv2J5kW81ZK
zZf?7Hk=q1Ho~0j4!eg`HXCv#PqX4v@QF`lHf
zpr8Yd>ym>`x4^i2;D)95y_a31>UtX3L(+u$#5Ii8^YIUXS@*)BL_q0yec``L
z=X~kPbTPTY@Pfi8J*mtm_-A109wFlfOEpu^dwM$Ra4#WETJzRy@m|Ct1z*AiAT*dK
z|E0E0jMbPQab?~!Iay17MLHZI1mA_@hkZ}~xjLhDtFHR>H`Yn0we1eG4PbNM;GISe
zbmL*4z2&Rk9KXc)+;LgzFp@Nd_GMg@ks#{^@J}|zq%we?_}(@1OL^Y#<{Wm+W#stiK$++ig6Ys^Z!Br3
zP~Oxw%;}4LUwiFiez+6tH(uImL&BSL#E&z9BDm7iEytY;O68^%aT*e&Fh+oj^W=0_@>mwtS>hP53r69-CpZuu;SH%a_&Qhm{U!ju6|vTufj)OX16w
z&S<{I>9uIG(+HVz5A@&TgWBbpQ0<=AE=P(|_92k!>o0vTwYZ~KW)sG$e7fDpXD@wZ
zML3QK;rqAfkG1L_;uS^fq_npYixG4OmGJYo!1ES8X{{epo`YR5kbL8zLitkOTaSY%
zCB;B4H81=_xks>BaNPpKFh;7X?tm3s^`A>y=cz{RT*0=Sp}QAu)lDQNB+y?ndU=7T
zZlVjD9Z^1Jm(&%7?#vp=!`kT!3fe|9CoR_&GGqNCvKg#Q0Tcn(Wg@u3ZaPGgC2SeYN9!yH9Z+fC=`^>|7xQJH$vpjm-`#At0cxW@2LU
z@ZmeW=!MwnmJs=e4<9>#J5rDg2o{rEA-iI^qLYo&Nmu#IZV1Bb)0ROsh_1w`-_
z^9N|h)VgQ)W%0T`qL1eT8XI`iwJ@=&_YP
z##Fm#H;5(WJf;C%wSTbg2RqH$XZdfOZ-Dyf>47=}d$#1>j@e-k$)7VBwGv>ne?5?s
zsdps4{9E>QK%%yxtDu#MFc8o;jyb?!>i3T~ey+*aT2rSwAzR|GfpotoAOtejfo$d3
z=%(edcw?1jpBXYMotIN|>I^Kz2nc|TP2+^|^D3y4enF(2xzlU+FpaOFP-;@&7M%(l
zwO!mtpI7k2?(rP%bwF_1=gL2qia!Gl%<3dtT$lE##zR;_S7$^%;XS#vVhd93w&i|_
zY)@@C0Bya3syY3FGv|x~Dt5xV8IUxZJR!7vzo5~D@nDXA+QJa-&rrj&Q+(vVUpCA)=cWxp
zF?_5PGF>SS!rp(0U9RzokEHiOWkq7*
z>wxX#f`$F1jz=Ke?_Gnk&db}?hGQdA&Oee}_!OgNth*IMB+cN7WkPY`f@haxrk&Bw
z*Zd&NC?!uRT3kgq4jy+BSRDg7A-7J-C+*`AtuPjMgQxkH*C?sY;kWA`y?~_xPpo0x
z(t%FxYoM2%37n>OOc0Rux)_viU(UbBFgCIM_v`JI+u_`;1BJyNn|EQExk55~my>yZ
zk3n~+@QsUtgLaPTiE%iDGYI%zrkJC~-;M_C0x>di;Y~h1i~{WU1NuNXLJj+aU0^K<
zk&pY^F^5HBEkRnPRvEIuP88sG>9Kl8dW8MSXG8(jn9aeLR=8#;#QM_di-RSEX!k$Q
zcdq3~P~dJNhn(|EYu1b4ZrLD1ysW~i+HZQB!SV|wrieJ|eYBVkPZT5%1RHEb&tJAv
zbr6KtUe>U+qKf4o(-!cXLMc3Aml%o}P~MnWeuU-2*Elec)1+i52eE>iqGzBwAk)2j
z<7_MGq^?vwe8?eXy}|ROf>0!wB!M+?CJCMIk#z5g>v}Cx5}32#t))I1xN~Wi$5@Siux3$Ibxq5X3RG
zWdh`qs-scsi;OF`BUsq0I>ySzcj8olp8lfXt_>Jj_jHvPG>~Y8wsK~^7r%(0XoiS~
z+4oJW72TDk12xBvsF`4{%aWd$7cZ5eDT@+AUe834p0U^wY2-8GBlY+9R`)@+;g}6G
zkhUXSVSdnEpMPL9Hq?EyAZ=Ods4puF7WVu*z(svSH^xU2t1?Tu(~mGcR|`Ju60HAq
z%n1qeL3@aF+s@#%pv|DP+`(Ig#cX$O!CcALv55|{pU`j$@81q{cL*bxLt)RE-Y$*8
z6j)dDU{48j|77&&Ul@A)AAH~6`sk6?6|t|Qru*3e^!;-RHr;kcK6z8d_B9Y9gy{07NThg9=OWp-?2OMQxc~q&nc_6uDp_(v|tSpmU=&k
z#MP&Pigu@=B!_ieTKFH$k1EPASNNA38h;we{?n5%jEBrK!hn+;HbrQY2iJ!|5h(UU485ZMfeWbkOA@OZM-KMWC8z%4CMkc3_7DnxEok+
z=GNQh!wbl#5sI)qU-p6kV8$;`pkH`5P21Sr`Cx#0rX~m`#rbO@nnNf
zJuG>OzUC4Qzf5FGco{4Ww&o1xIw4iG$TB89PYCOs+>g&d(C0
zGZfl1dJDA%GGeq!YJ_>cI)HqC=WB4f}
z#2xfoFFl3rbmkn?>>I*ltR$t;D&;dnr}i6L1SLVh;1En*&}1%~_!OYJPMq17CAjb{
z-<9LKeCnZ~<68|;XPHoL)W!rS8E-_m4X2jJ)b-533L5ZdDOgH(kE9$M`vj6Ma-kM?
z42N=*;vjMv+5*Ex?5|XvS~B|7h@1F0r79t>ssvawaDM6gWiMk*KvMwyxujhOzg{pd
zOb-TBkAMgJa(1a5rN0pg*j>v6jls9aH(^^LfqaJP0-2qd?slsu-MXT#dt*JVx7}Sx
zG4&+St2~ht1@jV)H3;^1JAe8T$7~OrlGMM&KJCJ;s0?n
z#m2BQvpYKR^(^U_p;CEzvy2*oFV_|?~uA;Z5Nn$%y8MC^W51r@LuHv++~EfChL0~c?=XhF2gCSs6>*gz%gp~#I!3ae347vZc4sr
z2o8xC_9n;mSk=Xp*jmCE%bV}z7gYW|7oI%I~EB!@oU
zgAR`Za``Z$+dpp+OQ~yw!$IXaR88|0*xl`uCY<1p>m3$x1%m}&ZH2*JewlLkU8fr&
zyX{gVHSdnYoxTk_uf09^hf2bb77geWB_P&P@hfk6FY
z_iY@*Am80@%but-XyE~JKZ8I@xdEtV$bOVff^M`UW2Kt)`-;o_DAK3F>!4uGqwsQ^@w{G2fOmzP@I7V{-Wja*Qk{sBOD|%9q^su8y2y8<_Nh||
zUccOaXlTd|tEcuad4rC)Z0v#NLE>>$G!qlAVh$Z45yeo@&{)o^$kxj_#*!I7bF!!o
z`PJ5TiGoxmQ{dsPn|FF}I0GJPaG(QRungGwu)zp5VuXB^KE@8$>^7|amyYR$b
za0Kb%6Tyd%9`#J>`4230{QNlyAF=`aNXY`;QupH?$BP!q29TML&Nx(nvmN^25R!F4
zO__Fqq1^F)EZAh3ppu#9I^n(3K*abfFt>i<$-pVZdKF&*qx)nO(dl+eRkA`O_X-GF
z4vDcNHSQa8I>ly^doy8uIod|GUKRL<_4xGzu;wl!g?(Ogf{9w?%{qf2)2hnf-3W;W
z?o^LVkZ`r@{o0*nt)`|XY&$Gy+7@0?QX+kFxUoHgI+=|!YK!2c9!2jAnJ3ZnnnE^Bp0|2CcITCIiPYvy_lkc0!XmDjNO|)
zQkpd)G9jayQJ>)15Ip{(ekxsY*mcFWutppLgnIX{T6wyDmnfOR(JS_(iN}|dp*4h1
zMtaHYK5X5Su9F&IpjT5`>fjR1JhzmTp_-MI)nxNC^HOlkh=|ean{KJqxl0sxo_#Y0
z=c^93SE_&@)pzi?%#aSV8~i5k(}0qIp230+ZFqr)^Lr{*R*S!rWeaMy
zhg0MN!EWup>+QS*TxJ(IfaJoHbb00~a4#93%|`*6-odpTt-Nv5)2>IL<_oy7y-!hb
zv54(3a~4>RE-WlGA43m*dk(CImhZ)nxsEuHL#wH)g9kQ&>{(b@$!-X9Ev(sYc#a4HcP!Sz^RpSG|TajL=70ZJY&O`1B}Y)HvO>YmN)QlfmJvbd!j
z21$?M&K&K6eBIK;FZbVS_0u;yKk&h>$kN1B4dW_W8f7PVp3Yy(~86r8i=637$ZO@(6qgJ-0Pc#A<
zs!39xKYwNjnJ;W%IG71jJ?#n35cCjsq&I#EU_MfE0suMkL7yBSo#20-l+-m&l^<=7
zuz+)OyrCN!ub-0%$K`>eeVO^v9-9lcpdL0^{~GLzXJun!6n9z(kv?7+!4{;UYWFgf
z?NA-ad=A31bMcINuHX&4q48pWUfM&sF!?2yJFDy5P(tpZ$p>hvmV-07jo@
zYT)=_6t|wR!^}@`0MP~CWBTL$mG($(;!hH8GoR&w(%#+>`b`W
z0&#;GCsz6lqhOO3}9B;3T)NS?aNgr+ESfccAS^k}D@Z4U$xJu;b2(j#+@?KIwR(wl|x#TSF
zHu<|cFaJ}1I9aGYSS>!%ociupP%#&~Hr+zR>PshTKiQ{kbVMMc0jCT>v5#4DXVh^(
z-?1YrE+0HqZ~X4Ie6ZK{ka+_s8mBBm$)o?QXA_T8HK_AB+IL>+M=ti_Nd)+ez)80}
z-BNMe+4|RK0;zd<@W70uUxfl^KrRjx7=c4@s$&)4s2QWz-`8mXi_+54jJ-ugq3qq)
z%cM8-E%|W|=eKxqi}x#wl~7!_Z-5S_IWP7`8G7JzJ$ZDB9;$(%6KZzG>rgd>+5_O*
zOqTeb^akYxhl@I>VRg`m+|U5Sz>wfy4ZmDU6KnA(%$gZkK^
zVsj!Y2Jg-8Is$q+NQ9M@^`j@6_7Aj^$M;^5PYp-4Jqk%5>hx7N^#!k#85IA_yDJ>$
z9|h4Slocs_zEx^HP`g~zzP#Se|
z0*H2PJ^}-;u^X$b|9x?$P@jtyI8KqxP!WJG5DW^;I%4N{B>`u!``OmBX_lI}))^Uo
z<%+h=v8WgwRK>8T3c4X8HGTKp1_pzmE66l07k3cUG(j!s>KFzA!k)SLJ7g8cN
zD?+6&f4J~gQB8rP$m#YaIh`WP&tI2yQn^fx{P})K&BsqL4tRH=RvYei@(uio8Q7Q#
zvCdzT7})Xdd>+3MLr_Zwo-XORovUBr08Xd^*)C&h5r}f4Fe5SH$xZ8%y%)oRW5*p-
z2!vhpvyPkeiHV6AOjww~xO2XKg>H!j$Tm>5d@rsy(i|QgffG8;?I>{4w{mu`yj*m;
zesG4Lm37YU^p77u23NK}5e2C0Ke2FIDIJIvK}i8$10HeR{OYx95&&&iM=QdG0nAtr
ze!Izg7g0|kjZ(C`o0(dduT-gz%|j7X%-BvNm^NnGqf~nU7I{~lXH$t+u$*&f9>12U
znlxVFmoeS4j1n%tMDMz!ULYX}*C@5pY)QCVF0~4EtpN--R}_e{9Z71lc><7xWl!3NG}HD-@4eY5%q41`coie{OqH4n7YLmT>W+2Q62zU?
zXIjHB!ctP>qDhY)KmK()iB83xjOj&tUKV8K`UB~%%(xgR5d7-TQfGU4rv3f<_kz>s
z%1{$hcB7)mQ>RWPh^Hs$UZ_*uu1@jQt5=!5|27acXA_;Dc5UY4U-=FFKJB6igtTUJQ03Y7!e5y$FQI$1t@HV8BX{*Bx}Obw9&
zp9SdYjnk0=0e54rV+qW~%j*!(!tlz%rWfvOP1Hd#$3ML~RT3_cl17w(;fD(pn|DQm
zvnnA(XV0d^j~1JUqya%l41pNqYB_!owQoZ(`z+fNlsuhVY84HYCIm=t2rutUpwXZ?
z;6rf$TJE{8hXc_8vMyymSr5>)M-b4w*>@MODMoVPr>E6(pB0`ycPS0MAyFo>o%lP!
zUqAtz8Xr%V4cOmW(gyQjWK8<<<%`GGB9Of`XcY)v8CMScM?QoBqWyq}AHo0RXk>xU
zB1A#(w7-eZ%DMXUaU_XyMa5A0t7(vqd3k$B@fZx3SmFRNxh|`d@i>Oxeo`j21E9?f
zAuAxBbTY6_BM{9L^zPlerGWyaHjsb-Wb6n&6ZPTKr_HUcSs`aOzyX_CkvoC1Uz@;
zF{Fco0{}K=Mn*<78eKd+dAvXQmi&`LGl*$19tz;_1lL&ceCK(p7|C2t)#xIAnG
z3V-wF&7VJi0DvcQYV)MFh4LCny`=wKUJiB+#%dH89)bBkdO3gX3?fvj%0MZulRuke
zEz!qMt4lev4HRnl8>-zmz+Mi`rAW>509q%uV*LboxFAsJv
zbNwXg`_!ybg-+dYr3x8u4bcCB7ltEL6aAO80{=$d=ba%)u4pLbV_b~Bouda
zK%pwQKUF>iN&~>>MkX8H7Mpbx>EaG{hf?Lq;ZI)-I(#QBH(UI1K&WE`yNz$$@Z3m`
z5ik$F&iCT?m-}EX`USN*qTc%|ecrXM>iw_53YFm}Msm{Vfn`v8n{fh|0sJL#h~4-!
z^(PCroiSG<&lRE6wkH5)|Ni~ks}1mhMLC9_Worg(%F?&)cU<~jfCdYpYrm?z5*_A+
zA;^^>o3%-v0(*FS_>nDsl6qZ>TFe@m~a<8xvIi85e_Vm&olTEHLME+
z^E@%B_rW$P8Chg6FmqtENmAb8TbgklWft8j02M(X+uz^cUK;3vCKu{t;p){9BJRog
zwC4jW@9fH84=l)Ca-fUjaF@pzhwxzeM&b8b}MxB13gqH07*BlxIIOe{uPhak=jsZR{T>F@&
z_Zx8V-CZ|`Jb~b43cj$CKp;l=c1h%+sKpG_itah0i=X)T!1Aj<*>Gw#cpxhavcH2f
zL}a<3v;hhl)I9p{gMvs&NVq_i2{5MNGTWtDfzP?)lY3gH-qO9Bk
zjv^q%wd$)|+y$v7F!7H()i)7{tfn7#ijLZaoK<@8xB8U4_bXv@42>kt08751VzpMs
z1T&FZ425}z`oe+V`Z_;k^@yI7-yqdw@Ht%mAs|gE7|a`TSYP!*%n@3pn2cOCT?AXm
zg+9FnyzA9sZx+q|IG}>l3?63@GxndOmMiw1h@X_CNi*7Ly4a|~I-l%cP5=GPab9J*
z-mEx?PK;l6LJ0t+J$b)~hX~iJ;0pvPb({28m}7zHwr7`+42(h0A0yopU@Jld-M;bU^e>mb=&IGHvbjXpRIvf&b{ry
zm*B$%uC=@Xah_wB{)Qa|j+{g~-TM1EAo3whArP)jLl0y@c${ty5KX`h5x!M_Y?7Zhf1*Jj
zt_7ssl=asK^rwm1=&^SxD3B46GsFv7NA_8bVNnx+O+u&=dW~*xujl2)$p&NFXBE_}
zXG&28v_@&a13-dHuk5iJKw?mNk|c~AR%QXS2aFw4;xMZSIdXZ;j7frJUuNj^o3y%1
z(pvc9`>?oW&Wq|<>~3^PvjFzWb7ZC}xfPxUAQr;E`jtGJ1D(bkC6*@LbqXB`Z3YJD
z9VUqiFP-@=PO+h7I=f_0bD|2Ig-Wa?47S2?O)oB9y2LNy9|pz*W`tBTHa6bxaG3Ap
z#*bDws<;p#>`@2Wi8dvR(z4f&Y1We`ZkzU?KsladzRS8EagHd*`2IN}P9v#ORKWyV
zM1snJS(UA>4wM5xd1(h=O``YqkW=gm_%MZ9@&i9eE|blFUaqV?A3IIKjW@<;GR6*$
zL2zI~K7w7sXF9L_IZ3MU0f(d)Mo?R=Y>VBE)h01_#KrgLlh8ciY$P#y<7BzGE0SHE
z6%>8*E1#DIzuJkX@BaPmxhxX!PzgYER)#_Kdurn37@G!Io){#Xpgddt^hdu~k=l}%
z-r4>0m!g37sJ@h@6AcMfjVO!yBwEU!)xI0}9qz;_oQ=^w7L0Ge0^DxJNy
z;KHo%8JJcX;EZfZUi_%>YVa6(F2+AbRXnAiE8vY+}bQ;=S|L#XD>rXON
z-($b5|4
zFHp^vapAMX@EFyuXvQhueq^GCdEWTHn0w2pEVu4o_*S-}DCkxJ0Rd4$y1PY$n^d|{
zy1Ux~5fHp7X;cu9lI~EXyQM)|x*N{CVE@mDGv4>Z`{j(6@r<$ee(o#QwbooSesj*>
z@-46)^MYqwzI-`ZD!ddH1gtgyhVw-1MoFe6Fjg>tkHy8e<~l)H6T?1V==f9vaM?=c
z00?=!>r52vu)W>gsrH019$pB3FX9q=Ks*56IwN>MZE^gdSDwPN!NHWqxqLZ_<@rO{
z?f@SyIhCz3B7iHn!Vn}Mg4ic*u$(}UMa*d?`R?som*c%zXJ90QvOmB}i(zb?ot*$<
zEOsR~gA>p1_&3Z*_%*+fFlg<+qZMV8;Lm~&2bkux6H%A9Lt1TJU0=4YH^?1eMDRcx
zo;$Fs&A`i8_tdVdnaBsXCwVwiAz6x)-|3E!$L>c~{Zi)@9dI?nq(QaXS28=}gUb$~
z0(@lcuU{YIE@V0EtsCkna@+4g^Km2-(40h*{u+6tEjPHJT|>c}BzVoaacb0hxY!n=
zGnwUEJH22$15ea|yQQ`5d?Z>ic0fbB5GIDZUZ3r~e+}x~}6G+0%1dK2fHjjYB%+F7n?&;@8
z4!xg)3|9daTwP=U0VA8xs9X#eW3p@$bC0g)
zR&Q7+0$?W@$GcL0X%Al86NVj^mN3M@x;)dF?4kPP$rBeB7X<~3?K(&mY_@QS#=Z_!
zgtgg=e)=(Co2Y|tdQC>+BbJ3$uKeV5k*a+VuUJk_%A7dHCiQBB)Y`-
zuVLj<_FnTi{#8OlJlXt&eKIe<%qlQkO9~Z+u2#sq_|na*h3Wp
zNN}S=O4R|yrIJk$8nGSY;dlJmk*!+=>3|?o{zADpJ|Q6?+AT;F(QU4boj{PVQoixN
z_ir==NLDi`jr=joF%KG<%ojIZb>6&O-em;WARS~dpk`thl&eBQU_b!75buyhVbAPo
zRCH4hd^Au%bwMPG)NI3^d4RhMY<7sH#aND(t9P}AG9X?gtzn-x2(3gAF+M5j2y9;P
zZtg?uM-H$uDY+?VJq7~lrpP+tAbffinWG`+NrS%aCa??9gt@2T!Th{CYVhD
zes|7O)~n8pZ_iY0Wf+o$N5kpj3C0(4+5fYt{kT7C}gv`ZXv~hg}4EgEi
zt;f6tbKS}ZY{6rE2o1mApF3MQO}zXngU%M+`s;RJsrp8$9c(~~Oq3lYXI{Vl2Mv>(
zuAJV7VBES@@n4hIq_;S506p4X9DralN$L7>u6nd5d3%B|P&Qq`r<eVYM
za4zVk10y3yXKe@-nd41u3=Rd6)0^k9~Tl*r8i12NU4!|GLgb;fb{W(XW%%JJ~<
zwl_AUU0XnY+5kb(>b-)hj`I-SJfI0~4GCkA55BM=aA;LF3@$-&T)&RuvvJF84TdNs
z!;hTiD&W0eyTbzsK)=0>S#oIdTnElF)oi(ei}wobXWK?R)*)rGzmhV2l~Y>qgQ&uX
zd*T@?3ozBz@uW6uf5g9tHbp_O5lk+uqoAcV3~I<}w%#?HZSyeczF2S@Ll!MZIk^6y
z1_9L=1~bx@AWVMqW~iO%OiR=?!Oad3d{R6=>HYJlVn2JoFLW}0p98+MlKJ?JfMWNl
z*CGAm$No%=W_#dw4Ng-`yc-?6p3FxU(JrvJpznsjR07UGlyHKmTl+}}qLO*-VWJuu
zBmqSVv&C^9W59)f1G5gv3=$H-B&XGJQ8Xa4vb^1}Mxeiu4QVs+6WkJXUiFJBhI4yp
z316}}bp$&(WwTNVwX#lh7p`S0!94@{fFXyHwc;MDrK+weH#KTXivR>T}%O;y=S
zpE&jDNX$D#+;_g&8}IEM%?75V#Cs^jaGQ4|$}~N@%8?3)?%zh9&GsxwAN&*8&tP6JAqw}*kpgAP4R>0&*S+V;k
zpsxI4WOK(xXl0p)1GODQ#DJ+!)c8r`;Np$|4hCit0nsbw?;ANjH81(Z)g
zMP#pVk)%So-31T!_CHI8h)bVaq>v0vJ6T$tonQD|o~UM<9k~e}yUbi*>7zdUo1U%E
zcDUIs-!62JVe9edmw!)CwpK}3hC1j}XA2AP{q@0HVlsub*y&g_0svrbeI0zVKKQm!
z42BZM^95!c$v=Sog8Ud2`svfBfZeUFu72dT|JrMYd*zDxaPb-_m9a`;MOly}c-H-`
zM=#u69A@C-SPYwP0fsU=TLd{I*eAoocje;+*o|6J!RLb#dI8tm(2;Yf;p(xf*i6op
zB3qBtgS$XNVrYo!$!_1~F>1LTP|F6fNi^q)G2OH_SD89wz3PI=VN3hr
zk`P?K{-n}l4`4SC_A2n6U?a7H4nsUvKo_&TPmUoJ=d4@^l^22}hAW8JNsxk
z8Z=@BSk*Id?Wzuz%b13nH2+0x;?S70R50kUn1bUrz3`N;YOlr7|dKlrAOLHl_f!#!O9Z1hNerp(iIX8VnePF;}qG0HOhs(J)8$i5GEEa=xI!xWa^5
zI^8Ljf%2H_Z077Vet>GB!(Szxvz1Az*xg)H-n#UUA-Xd{WGLQK^qTN9r+eVIQBqT1
zyl^3k^dBY|<4r+iW_!ire}jH&
zlno=SU0t9*AmmA77Jz$Tq{1E`87`kidWtr({>t7HEg!(p86B{lC1;yi%PauiFI{ZF3IMN_TZ3Z0t9DNCz13OiQi}P7(SG$hB-=3=Q
zuZXsxD>mAD?TPWq94^~lQ8(wo3J@(NCwS!`4HtT;UAm-vMAA9B@kjS{u`u|4|X`rnwO_O9X?p`C)
z!q&ADQiZo;>VP2?v-jU3z6xp_P^8aStS_M&mOr=G-n!k@E5?Pc5;z{zXYS=b1&NlK
zRSAf%0){65*<$dLz$&Uja14o!;_@pHLgl1bo$NPyqxs^~AGY@Dw;)8fJ_l`p7(<@H
zV}CvcQ5Z5ZGEL}f*Mh+jD~A*o?8@(I3|W7#ph6fAx$F#&^b^ohA;F$M2z8E2kOU|4
zgbM+p5NL6l^`ycM1CuGEZ5RSYJYEn5K}ZSlEaF7B?OZ@LFx?B*V%n0ko^){!$T5uf
z{5_0RbT-_9Sq9t>79KJ|F-d@)IKqa3-~pnqcR*YWPzsVBV$wDP#zo+823>AFnBQpT
zC~68Bge_9bo|N}TWi_t;nxPuRmZKUd+ukF3=0y@nWe^cN
zwQTqgL{~x(kE9i&g}o~g<(*r(!7r~G%ih2_Gp-Uc|*r$_!bAej;L?wwTC0A!aKX(=gx
zA;_!xfbhaeYNf?%TaToZ`X9=#{u|mX3@yqyK4)kNjD%cDP^*z96uc!t6%+#e+-=UK
zp{^I`uT{0ha1pNT&S5W|t~lx@fJKl(wULyRgeoG{Oby58i@@9gI6PA82D!xp2!+p{
zFxa=)3Wbl64x?Yxl7kc>Lj@CR&r?=m2>v+a(~%Gv%6v*8@d73cse-_|CJ+)9<^awR
zY&rQjzAZ4quyjxXXDA{f0^%9X4=LCBM$lE=q(^qDnL#U(b`9(8a|4{sQ|zZ3p=AQ*
z0%BvJN{LBggg!aQ<3+S0`6y0vr)Ft9zZ&Fjd#cCv?hY$|k}1)`LOn$jb6-!V-^UaL
z=es@?>X~Xy|KDHaVGhIT9znP_Pqzz-ombNY(QMLo;B=&kb#wKMZsjByxIUJRQ!TZz
zIz=a|WdMMN+NlZ5ot<&8zYu;7<{8U_>V>cp%En;*A>L?h!b0&r8QwE@#da=gpfJ^s
zqU!a>M%$1pQDY(3Bv20?~Xtx3vx$Li@nCZ&$asq4EN$483x29yQ!mAk0ia
zdhpq{Jg2cN{3l&BQZb8C|5b5|7$B}-@VI?_eW4}-34zj>!2>Qb>4e1SJvzE39(w?|
z!B2#8KuYMtyWmQNhWzuZBuEhV3=Q#58-S2N#JA{lfK;k|dyhgD*?$(@Ilmo%+1~B7
zEH6(uQxD1p#T~1V$r9e5`%-kk0tv0>S^)AO`DIw2mk`<{Qx
zcP#jgxk-@1FZ*^MRn53G&%W`2eGFatwOXKAXlK1Ki~bn4JcAxA6hS~TOt2C5iz@~e
z{r*!^QvlOVK&l7>*@Ry3|6w{)w1NTxVuUU`
z1r)Gw>uMKS#Zd;-u0vJiS0P=HT}Veu?gLV?A~-K=#biVxOz4tw3ANMyfsCC__Xyo(
zDE%XmYDGPY9oo-SY*rJ=TpvLvef2pa
z0VLtpx2tY+EEU$=IwYQ$lpH99Pi
z1qD)}A3xr{b<4o%dNHU;87-~Yq|=N(myY92poW5pKvtEPci5{m2~Wj^M%kO_Fx?!iGntRfhA@O9<8T|ki_
zdM!uaZ^SAaumdovaU~?V(Sb*FQ0t?Qk&0yU+(-ssge0AX3rzM*I(XKK7ggmY{nn0!
z`1tr*T_Fq@_g{MK9V|zawBQg(wFAc9oN{+*nq*R&E=4|toS_G#rJ(T4#G?cdA&B{8TrS_KW$kj)!O?j3^Yvv~U6phdONhj=E?&J2
zZWqYRl9N01MJE9wRRtc?;}{7tR|9a$0URlFEveXCnK4ir7AA_}ds=9rjvgwqmbX6!
z%pQC7a-s}AQ%6S!lD9|@5;Q#5;D3__3Xb{!FnaRoA=CT?
z)HBI<@7Crz5zlskzCy!#^CmSmB(&9)p^75&q)g4>f1`ynzOdFk{rkkw4Nwk9MRw)B8EpP*Lc7nJ_F$F;pm@C~iaQ5n1Igag>5dvU
z`vS&q4duXf#lB!!p^)qz>|xNxrpJ}%QL#luKN5-8yruTL(;qHvO(LFv$or>-8NRq>
z_IdN&j@{>-Oo2MnhaXomLvzRK)rbz%I&{UZLEah<52;`T01&bR%xGi`P^Z_$k*<*3
z1h8@3r?NM%U-OXP-Vv+A!3FUpfsW`_BXelgheIn2Tiy*oi7@QCz>DCpnpRGO_<}>g
z{T?h-EZ!kGVEgglr-RhMD>fjJ1(71+g+uTRRZ6LeN;k2s==n6dQ&Rd96{vvdH2q=nN`(0aA(Yc2BL)>^Gx5?YFGB}>R(3aT
z99(r3-+mLQ_V@+WoUUMo!Gsagt1#TXI}SM(t{7Pbh?Ah^$FGuD_3I#x?B@{zEDSt@
zAUdRQPY9y+FeBJF$5qE$*`OMTy!}8R4$(&8kV?<8OZ`EAA+#MmxS*NxW3cRYDRg?X
z={92a*QNJ%hxh+Bd~l=Q;7T4Xveu1eheTK)lvS;*!2oC>SBSJS5aIjvY_E(Zbe5Gc
zacVKtyt$a(%Lm;P5c>Q1`9b{_AE2a(XZEjT;KEslc;qi{_+vi~&CdQ9MKyF$>)_;%r=w_r$@ghD{tP_X-k?kZq#l
z;6O4NZ>|tl7U@7QoxhOB*IoA$_21LgcB
zF*}9jlOrL9Bl@2v-V<|Mc{GzgK%oYFYHUA0ZTDMJr8**5oeI^hGI6>Kvgy1&qa4Ra
zy*&m9JvggKJWEMA1%4~|1_S|yfYt#r$3dJA=|99hgZi`4iUT+d=gzHLP<_J-u`B2m
zJQ&9pXR*FtCO1(c8fB$&?z_Vp=za#*D+~#f)0V7AoX|i9!N)e#FT#w}SrviYe+Qia
zokH48TyHR-NMRSu0(JoU$;ZbBx=k)sF0+%Am83V
z-n;YG!Ck#xF8zo-yBh;F^0E;XBy6Wtu=^W%jJ@7b)L=RW*6goDte=Hdx%4==sE}?d
zt@4@vlu%_Sw0ZjbRQCL=vrz{EwRfN
zTVcp{vZ;g+gR(7pHeqb&EV>?jzyaZ_JlrV)jyjHe<)&-GJE%h*-~GePWpH}GI#I4&*oH8g+k|I
zOUlC_@=?~a#}dF}ZM^xhnOqKk3!pEr=d>RPsE%r#1-U~=L6Oz+pTepp@7;H{hvqSt
z4Tn@YxgpT;(2He3^W^#P{
z5rcbH33KNK^5FYrM=Ser8`QcH`6DS-uY$g`;2|JW>`UawPl^?FZ;mUTb}3(=YS+H5
zRQ(@7LkrX3Q|7O^quAMwlt&u6#&`=4PMkX((_QRO@U6?!e|vu$^Sd@sn~ENvBdxTs
zM3fC`iUAo?5YL_2Xs2fKjo_rCh^F8SR~l8Oy!#-4cl}Q9z3(=R{bMLnx|E_yyCy>E
zn<&!cjfx6F#&o&9O}lc^m!w{?*}H3RY5i1iMcN*L6b=6jB6-TFvzkjIx&O1wJY_2`
zoyO4)Z5U5_%vm1yQEm5`aJC8SAz}%SHSTT~0
zUzo8z-NRE_i0H-~Y{+YlovWJ|5?!g<7RR?o0LQz2U*Izb)F1xbnpyI^ku9xZX}b1H
zHG8Vn+=&wTQVZl(k^t{%<&&yyNxsaGidQq776zxP5M{?Fp9j#)G9+n)?1erUgw_vD
z&wBGvR1Lg
zgVPelQI&f7QlFE0Xu51rVbYdz;Trz?9UX1qTF~4tj}2%zSZkEz8<>6ray$vW6f&U?
z##Bp6Tg4x&`?IZ2KX=@l)$u$k7q{Mh#XVb-cHZ}pT3LH{&&2xd#ip##XP4_A$G>#{
zF|pq(`*s&&Q(k$qT6k`Q8?vdKY$hN7*^HZY8z7GR~kWh(svBblSBty5ab1|NOdsz8_rUwoJBKP)%##*>nzB1m#N5^bFG6?>(G!%0>$;JFK
zae#~(Pe}%k6{1U1AKsVp%CW*xL_{^jk_DN>kL2DCG9L06u?RxJSsC?Pi!1-FKo(*V
zGN%5BYx)r~;FTr$6YtOMMiiKxEnmYWaWyz0L6vP9fNf0AgOo#~N
zmACJ5ix!@t(z^3H?ItSZ!sfjG(ZTDaS8gUwZyJSV+4vF7XOoTa
zg&V~PX+s~~JJl?*O5K(dzM|~xvz2!jhfOWxxSHtMkzV)Q+jDn4u<4qRiPc@WwVD>*
zS}m})wv9K@_nHT-yHGM^Rl(=Xb-H}xcltmtZfq_X3x1dH5_TIO>4+jpiH<{HFB@X;
z2C6A_J(CCWsHmFr@^(HCNWVHfYCYXQJcyF{II(N1xua7zaVf}q;tSt3VT-eOaw2Q@
zRaz?7&v~=hRd$T-CP$?4
zpl!6EY<4~rgO~bepJwd9=(hT-S{Pde*O+UmhDtK)8fj
zW(QfVKGO3p$?n(nqaszDE3Y@sbg4Pp^G6c!!OOu^(it1GQc{?HdRIcPLZ0fHnng*t
zxB95Q=Q67J+Iv#z<-TSfkCFA|MDYTk06N`%A!t}apL94evB3QYJC{ZmF=zAEel@l1
z*#{d$#0ojyqor+Jyo7;_N+1>#@`&I2p=`lLlAU${#q0Cs84*9k%AHQA=L6r#rn6el
zry3mN7}z?M2_E&|8h8(EYIx9z*D8CSkWEKLN6{H`VGPUN%}s-WAWsoB5MMiC^D*v+
z4D^cd9ulF(HAYFecAYQfH9U{pQa^G0S6CVMa!w3FRx}*`Uv~)Hr6b48krei{H5~A!
zn>X(__V;r`Cd{b*MKhFUK}8)@5Jn>jOMLuk8=edmb7*~FWJ=)GPgiQ?-_K3TdFL`U
z)XAthGnbGvAg-mQ1t=pxf&-%CY__(xki^75m!P1a{r;|O(eu24&tJX({^h?GEw#SA
z-4;p{JYY-NXjZ`X`J9c8+zIQ_2Tqk2MH(l0YQD94jrTr`)2XCGQj~Gzk72>kB%i}Z
zP$h$Dz+dWXfEJ{(@I0b+JVg6N%$1#RE;mkzBnucp9O0{8B8{N(EJ#!
ziqCP>*)HrI$}I!S*x
zDt@wE5`D)bkwr{u9j7ATbGpd(ciyKKjvQtk-Z!nHbCr!={jGl{T`giDKe0Cs+Iq@k
zBBkFwjdx{?DCY+bs7q$(xqomf*)SEnZ9?iIwd-nnPjHLOP>
z92s^uVWpJ8w7YTa)F&5-5JC2Ei__HWmG8%OYc07e6>MeiPV0<)GF~BeH;{^zwa~DA
zxNrgEwQ9Py80;x1*xH8dB?BTvdFabI-&Xz8q=Qbu|5N0k)2VmnAfnVS!KfO1HrV@A
zM%fgh@ZNjrM0|;0@6-oG889OI&`SKoEp>OU5?l?&be{c^G{^sbBW*Gu=Wa?A4s;zv5gBVSt#lYchI#cD6n97}N>jS4sJ<
zm$k==V{GqfB9Bu?0*Oz(Jyb3a7%Llh;S-Db!P?(Oe_jb#I~=AJ!MUNQ9zzDA!-wnE
zCNDF_ujNxoYM8lZH8t7RzWdJ-t|RN1sa1mMNG*}yq(bv6A8ih8{N*0UM$D{@al8@W
zkQp?zv^lc;384~{aY~Kzj4G{Of<wSg*wp^#@0Lv3y|K+DW+er5^omOT%;hzg-RIx5OEy%PLd4V~V5C^GAxu*H4m
zwcsM*!2G_Wd$S5zB(vn;7rIb$#$&6m|E-9H|3k=OwHw*LexrG9G1NisN5vKBE^cYB
zuU`ZV6C%f7;2+wRpTlAp%f#GDA71jTJTfp``z(HD91S#K=u^-y0%X`!1BRYI*2NRN
z`!k-NXA8YnkwX-hMiyKu<7%x!IV1$}
zxJmQPGR~H4F&G)y{4j5QcP0ZXH)UqOqGD6~?@~!J3l2n95VPzPOT#2*IjAO=fN@Z}
zR(UJm<%q7mS|m>qxxnfLCk&0nsqAM>foEA#GhIiW{&fA7jR;zOCp!4OS{wYmkeaKc
z->j>`Z)Iy`=6_xZ>o98l&r1Fa0%eEcwdXL!fLkock)jpJceH;dFE-3z*YDQ<&POv>
zS=@Qwqe(-KJGPs4>PDitA@C4sv)TPVSg5S`Imz`eThZKGQr*V?X4Gd0yw|(f#UARJ
z8gPP17eO!S+`;_5wQ_#yKpG4S<&4buu(n6JMnRV@)Cy#EwW~+SowTGp#{KWZ-hr?c
z&Df+G%O9-=B+)nZ%b2@WJ^e0*efIlu0P;JRTRs63vWI53daE?TN%;HDG@+(G2+Xp1
z=e>%fLfhYpdCUSmt3wxZ6?6gmLoL(5R}c5J3#Xf_^4`IWb~W@~;5#YRIzBc^LDp=T
z6ohekZr4_DBM|AU&72)=$!;u_+jw6~_ccd4hiB=T#N6?%y?d@WAiJt>uJhD6bgm^3_NJBh
zYCyyxDqv?E;=#7Kzg^-^Wl3l^N7tN4Z}rUo5f@$rxfvK6^G-LEuC0^O$egj>t16r4
z$7i-W6_>jbdN6SV!&?hb%}U6G>?G0(0_Nkv$qBz!Ib`ytlwdCy7<0vdz*#3G<Wg*Rd%EK;=`$N-$A$cw*%-RDR*zZC^<4<)
zz~iiY|KEpC=G>wQTCMYx_%`yG~iZNV#j#6C|
z%5%}?o!s<-&0(yk7xZ|5EIyj3ms%1g-HsJMTJD2ZbY_4gI!n#Amep=MY^*d?zuLN`V8z;y{
zmlW<4nC}efuIwxvdFi?@y8%TlC8BWNcMewBH+ki46
zfV=kp;GzV=yXRfz{!Nf=#K@(0`Fu}KV9Xv){cv_gG3CpPD6%HrW7mX|1awpA*wH!N
z^1S@#8#6}zh+QAgzRrPKebg3gtU+B)-ALPfee9Z_=Rl69d!9Km9Z#+zZ)h)d+?MdW
z9gwYF=X~T~v$H%yfZUu|e2Ez4@-ZwdWBDj!TMM1A5dW=PmyD%)5d*)XZjp#$q2%rM
z9o&YNe?-wxSw0OIJ3d~U6l3`B-E7FaZ@apB54*`v`3qw3zP0_7h!oiH_&&gid|t5t
z^77iwGGCnn`|zZzZ%!bLQoWJIR(210eq>PMIUflgfvMh)$tPi9%nO=Jl`gJ790}?W
z-wri>9bUh2wB1L@dYPHHxtq9x->R2XCA|b6dG-Sem>6pQ!|4dRFMm9yLN-meo(5a^
zU%SLohS+Bmm3RlLTBe?x>}qL>Y6!b@SJ899vLhZ^j-JJ6R%2N%Zt{p%B{X5v4xKmzEW`0*Qm(cD!s_<
zXuFcq&jYW5t8C4a`=jioAHEpg6y^JG|?pIrt3H0f`!)Ie7g}b{VW6CN9P6~#OU*v3$FzROw9sWxT
zF&b(-d^ho0Vptr1(N1Y)4z{YKT~luE&{SJ
z^FMhfX!7iMRx|`v6#Vi&Z20&Cji8;~1dL7OjD_1Eh*SCL~xpQJ`>K
zjvRwX%^kutGDi7zKXPwR-wl;UX8cKElX(YH84HL
z7L)u#Lg%=DOy@y%)pKMT!qi-k*b|}1H7oHss84&Kr6`qeyV;E9vFMtgbr
zj}1@xN$jA7kgaV*yD)d(Xz%yU$Wb^oK)$3>%HP@B89%6IV!S^)(v~l-?%|tVDy`?Y
zQ!w^_Wd&B*zXt~g*XpsB;Uo^zXN$fn18TW4p#Ybci^!ivUb+weG?c?kO?}iBr_D7m
z@)h0;SphPw*j#~h?!!7kUON+nZ9F`j8Y-rz$DY6W#k(a1RlJ;K{}cnyH~lGzn2{4c
z=GMReLPfQ)6TV1^YUitqDAk{{Q;n-Yx@0xR)JT1Ty=5C`>-_#sl%bP0^4T(qaFMXE
zu>4%-_}_o#78VvJ6_*ouaU#ph1%Tc_S64Uh3N5~nAKshIrG-52sw9~=!(ZC1J+da}
zI!iFzC*;+fPXfH1U2c|*EZgw#q>TP@)GyG;+x`^A<&|s4}3d}!{&2n0aymalF^U9S787r%-&TA}p@7{$L{YATRE;#uMQnsV6
za?XoSg30*Yl)6ln|0Yuj>X`C-l_!bh(bH?gC2}d<_4PFO?)_ZU80U(LisFh1f|s(f
zvj=^83x%(bBz0<_mL@sbR;~gKZS8JurIfzo
z)_Sa{WAi1A5!aNq*bfep1!aF@M6)@Fg+~{^A@k*e3dSRbHaVg^Y1ETns=B7*B8JA&
zgUMeMOQ2Qbp8JH9HF_%BOj4t^V)Wm?RNb+E(U%1f0YM+es9Z!Mmlc{;dwN
zruNAEM-_agy>sYJ@}tq~94+yJLyI?1E8$UTUPqg)+#pDS
zoeL{|fjYZyN*&qDSdd4NHIl%ylBqbA;b6bX$LPIWyv{)8FJwAv1%rj0XV=8POFUPd
zz~k~nyUeVe3>}42qPkI}yppp`qU{_v@aFV|&)}1ojArE=7rw$#;mNe$A|qn0hOH*K
zOZUz4uAJ(y0he+rpL<+ZJ(fj8v^A`~#j@u1)@D0PMQ7iAObklsq`>3@pQD1=wj{D3#Yfu)vwqB58UFZM_>vY-pv945f+c<;6t3<`$)sV=0-isW8J>
z>dJY3Y6_@Nw{UuXdIb$!ah~5(qr#1rR~@5WDaN7iRDFL#SaGbISMt~!yg5`ERe2!}hM
zd95GF`d0|`R0h3h|?{@4;Ea9kzH|T_={mPC9mOLna=L|
zeq^O)QhGMm+2`tjFs|yP$Pc-1ZjWrnQI>fmbYPnvksVnPASa)JS0XNR@v=*re*L=D
z^0TC?B{;XAMUe~6bnb1SiICsJnA|u}qbB`KnkiPt-ghFxvKlJA0?7AO53tEt%IJ_w
z@KK^zE%l7TDA5KU(FK*`5;j4S#3v2XpG|
zWHX-0auXgrKyY1X|Cg<*ZdEx0TC?ZR8-C2Jhi(-(=gp_~=>vFfI)7TmQg(&b6SMEj
zcS(&1o%ky!P^w(M!d}L3I*zMA4L_aGy;m=$pB=sDo5R&Vd$KNkxEthm*usbdmx$Cp
z>Oh6r@9mZG3w4?tRRtCVzf3SI)su>Ga!^5D^9&E9@D9K8J#*a+cPpMkC5&POk!nT9
z2C}I#GhnE8<$9uN|FfzCx16Et%WA^ChAiJJXE;T(z@HEH<+*HYNX>?YrP$
zVxDw$07qkg?qz)?&8h$JObDiQx~0UkzG!64qW@JMqn%XKs=olA=hKyXYt}L*vQ=uRK1Ae!Bfnk6zUhmEriK=Pg%h6mR5?0&A
z;-{N>$YaJ!;EHzp{d^rSb;XpX2^Qjr)PDZ0QO||u(wv+bmD6{5zW`~Z#Pc(1?_+PD
z>!JNZA-?5nF4cooK~S09@x(p=(R@Btbp|0m^}kD%Er;ls+9Zi`@i%jIEdI^U+u*M-
zYCZG8o(Oe(Ik$=;B%x>3QamK-*W}mj*EY7YMTTNS1QYYIX`&-T-KQ&u_tUMC$aD5@SC}U_Fb^)Z}zX<+5b!vsgb7kw28Z$?%CHL8d=ZbcO)Qs-UPu
zJ5V1f7-g9&l1#e|ib!zj9S;
z8eHjjX}N7V_~`6?cfP=m_x0Kp74tKje5}tX@70Ze605_X(m`U?AwizzYHz=A`u+6c
z!50%{l9Ek#Z3@h$rHEAiZZB+g+sNi%zmbcELUhT~jp>2~Qk2|so1WDyOfH^-*lWk1
zXyGG^^eZYLJGdrnkb%TYs1;Fo)-82l4qV+Hf}?3Idf)wn(S96zC)JfJlZ)}
zaJCD}aRiBdx;{jFjnsMfU{%Y%pfc{ylA`z1n`l9|xmVYe(j(IFkv$?pn<#-b4h((%
zbwS$g;XL@)Z>B_%F4B8H4NZN}`tehC74?~X#=$Za_ZC{h(Xfq;bu4`7=HzootgEN&brYrre!x`gf?m2M4h;
ziJPnr=Zp2=&5wh$^r_e$nhPxQ=S>eNm-Y1Mb?OJt_v@Z1Eu~AFCaA2IU}3YlC!RZT
zIWwzUG59xx876!4a6*;J3iSu`%i{&HWElTN-rj@--tcW`VbK+k9;M
zNnKbT&c;^n$$_^G!LHVrJ6)Fe{aGvFB}fi^{CZQcnS9u&~z`*KL9E
z!Uf*eEF^44HQf@U@o9!86PN%JuG{8L*{w<0&wGzf
z;32bVx^X8XB>|kNcW+sWq)s15E{e6P>a~xS7fjw-0P9!F*Sa^g@m`hf*7D>R(Vc~@
ze&veN5-Q}2d{Q8iJ06Z=cbm9SaZ!?AWG)~RCHip_`&qk?};GQPOyw9-fFJU8kd6h@)j
z2U3^g6_@-&wZ9M!43v#I6XWqeASJg#qS6{eIc+~f)BS|z$bQ(@DZv*DdK1Qb^4+x7F}T|4ST)()%{gL-TJ**DUWnbKeNG!
zw?^0MOs6D!vn;JVk@r4nyUvmLCueADG8kx+)k1FNy2`%Y@c3n;)mt~WC5ub`*1Kzc
z?hDRk*0ZU95P6CC$@4q>Uw=KKqNbxoo0?*@6q%-)`C8X|z76SkIT-BLR~Y1cRHH(k
z9%lUy?y4DAOkZ|C_t*p0r$a@ABVK9+n8Q-?%8fMlW3Q1_YC~Oe+|O4?#+^I3KKWX&qnD!l#YhnLex$jSBm-?1g1ioohacrwAjz~FaLH1K40^<|0|
zrKIX$_-N8Lxc?5WZ?8VD;rJBk;n7t21bO@_MBNW&6%^Y1t~5sg!OBCDJp6o7Zn#=+
zYc%<*s32QQ!UJMh9ijtD+`%}CBm!6bQ$>HFfbOnA}n{6@(xaL3
zL6g8hrK)&1)!1=`Nw|uQ8DYCZ(UHN&7xpM+Y9uF7GHQRjFY(Sw`1Y`aj>-z9<8hyD
zy6o{|eTRIv%?`nxRCCq6gGkFC$bC<8{^h9N;JGtcxOCWZ#k`-6IlE&!W4vi0gOt#~
z;aEd3*r>L6(7ab+XZdfh^hjx^H!@6A3+Vk|+x4WrjHe110>-Yj$v0elo*uzUmCw;~
zv+Zndo{_z6pH76BK8s?(d6Si%y}8K@5xJSCdv&$y67e&-GHu~w7q6K&Lv#R
ziQi(*TOVx7WM|*0=$3ngKird_=C%VQDQ07pSFfBJNUa;9SS>b(#SzT?7%GjNhx7Aq
zxCR;fT7@zR2i;*It_^VW<(&*k|ImB7&{uk31wQEhR&${dGGY-{5YvpQDU+l94R&UW
z5e-<3UhPl(m>$=Xyk;WzM?Yj87rx1ofO+6wQIe7A6g&OU_?L<2k{PI`J&Hpx8G9S!66Q6@*@@6UB{Wu`0
ztn$Ek=`j39Ru`&>%(ZK(bEl9`rerlQ0S7cV(ePloAF?n}t$KT!S$m+aTy7WT4#
z_Kc5qh^|x6J^%Z`DziMVj?^!EDPo|~ymuk%uWdnmvBuM<-kx`nn-a_K!_e&Oysg_E
zW~S8g@{*xW*O14~K0IcrBZ{4ux2f@i;E+*clLHSW?W3q;LSLFkbAiEkFdl~2RDC;d
zCk{U|Bx=6k{9dV4yzL=BHXV)RK%HG`+upp#)W6C3dIT$*Ypp?+rb{CugXx+E0R-Ia
zwTN{=FZnI{S<2)3egKJ((nGX&B8hq@)HVvM1K@u*Ql%k;-|;FadJSY
zby-;7hB>9XG5N3T`uaLgiVijN%Hd=vu=7C2UrBAXcHjXR!5E&vlr;3--X7VyG!Jx5
z_xJZN()rkBf5yfoTbD>ozX(9i{+lZ+fZ+*F7++o1i(5^xo*DwmCo
zjn>S6tRf>LjdRpmvd%A0lqv!9gWur|yh{N8{!>k&wtKVT=G4p#bla&Y*pgZM+fj0I
zaz1$Q79IeU2yFX*{mb+@%Y5EJ{lglNhssMzB4;PCUgNd>^9!2%;f%+K4+``em3iikTy*9nk*A#DP?Ym?Bp
z1dZE39AAcXS;p#@q@OoT`=v_Ht%4|#2S5cX2(Qh>0~(-eAj!zh&7}k$r+6Sf29_PD
zj}N7hheHO;=EE*4IZzcVA!KO^bQeV^fI1b)69LZKYX+=QP0)V`y+Vb+A_XnYxMTvm
zUG;XV7n-`lwK9j*UaY@Q{KUX>=RNeAv;!R|71eYuo(J3xuLCk~R^V2Evl`)5a?u>7
z(76O%5pZN8bF$eBob+Fje}M#pssg&M#82vPT)#eExbK6*%*e<{LGiU}FMSobfteF}6=r53O9f2s)4-L7kTDfm48W<0#H6I!I9uoOmHe5N(ZGPQwY8NAUz(kr
zJ#+_aqmI%(U5P}Y4VOSF51kZ?%A6@)~wTofC=cTe#Y~lUF)UOyE
zq19j!O823k19K5S!WwK>(R!(HlZtQIe!e2eVY4X7FtV19)2jQ$Q
zFE7_4Gub361IpSf1@(1x$jJ-5cE5te&lmrcE`Z@fM>Evx2eFPRL(5Q^;{pK@k-En8
zd3*)v%K|-D8i%+%si#Up`oFKj2fHJA_?T%qvc&FYx{m}XM9U&>VZz4$R3bu={wD^
zP|~V)$U>E);Gt5ovf;M~2vkx91q8GRg;>t9#I3nP6Y5nmfnwkU0l5Y~hIr^`hd#2C
zHvW|!V6#rKIhQ;O)E;n9!*A%14yRLIhSLV57aO(lSyb1uRje|62%o*bC=aR^rvUO2
zXsVc8Jf9(858H&6=y;%GgYI(>5TaYRdcHre0bX(7VH`Hf18P8HBf?8ADY3D!u(hDS
z4#=5OfF1?T3(@Et#Sym!Q6HVw0Z~84EE0UN9ayD2p+gKJNE2RrrKvBIkQdYDmeZG7Y+7N-+yQ{0~;lqbOT;Tz1C@0fNw!_Nj6vM{N
zrt5HAAm`T(@|g^oEy5C-o7OCPu{aQ{Imy7Q?V{+4ZTvZa!P#Y-GE{dmy>IQ
zmNnovHvh8YC|Vb?0!{Ag&o9{!mW>Djojc@NKp|9s077FU)A1PNu=0ysr9bK6B=qB?
z;LE8(uPK$~@9#irR!nTkH3Yr?o*o`r?GsqfK37*)LyH-lN2?F5a&NexTTV^^T5eiX
zudK>JTjNNsp)|rZ3C+LIWTf4f^M5?L7CPK2fW;MhMoTn+8BYlqIzdrDa&3fDxff2c
z`4Aos_2%jDzi3s4C${(%O75F;6bMZboVF3#sh2l^ba4SW#Xs2C2;o!g!*ka828}n+
z(47dZ6t3d8Vmm&R(LKweFF~YRr&=Lj^@dbCaCgIbM6f>aU=sR8ZKy3co(xtS9uLik2}@6)(+a2uZ#RPw+|-^p+uJf1=Q}|)a03xO
z@V^i%E&@-+2+S8k&}17wej(8UsBu7Hg4J*WazEIIz%wZW+n9vIgymVa4{$8zrDK)X
zgJNf1bC$NX0m-ey|KF1PpYdyfGGFBH|C;UhX=Z^$Qgg!U&JGY*t;IQy@`9*=Rg2M1
zgHz_6z7#@_nYLRyLe1?C^b70l+w*W}OinwH_U{-U0CQ38?^);=reYd$`2c;3$ILsx
zG?}Nyfb3BJDmX{NX9ys2)bM>~W^3sAgid2<8`A<|gO*KWC-KNnu(G0}q8Kk=Dh8GQ
znIGDJff<3M8Hk9g<^mRnidH~tW00dzAT_E$4^OQAf;>?Qp{gc7%tl&IYiptTQj!ns
zHhchbBF~MgV_?j&&QXI_(m$0L=Gw8bDq!!%tZ!_9X3=u+gB|cSC55CJ7zf1Q_@how
zD6I{8Gh|iPs-^}MHo8t>58ktRal=w%z3yEEAADK?n!d?gr#^gM0xdTO+??0v+>H-rQjm|RktH}Xd1*ZoUbqIkYz(Mei45F_m(w9RY
zqfl4E(2qowYyn0^jmts1CqhS&b2#EW7Nyx@IN$N9Qa#{@7~pZ{7M0bG;^N+Nd^CqB!|X00#ecnXt@3?
zDyE{+2Tk8py}(ZjO7yWlw15YlBLq&4{(6m`TcO3&Zz;R{t=v#Ug&ef}M*3@zch1maU?(E;8PWL!&4`s`yz1C4f+U+{!
zq;{)mo05v+lwvK)mN1?pLLNJ@rD?h3l7~v09<)MQ+RjdF9!s`X#xBC3q*aO%D-{=x
zqUOB5_8-_E&;01R%*^+`KlkVJzVEp|ulLmJx>fs)Kj1(5l;^9rI=br8J&5Iq8PF+B
zZ383aYE*Q{29CJYLYCoK1`TLp!fDn7N#t
z*-%>xf@=&{D*OBa!MO`23KZ9NDuDo!EFD`%OduHzv2ZJct|kAscl&yJ?J-mk&7K`3
z0hd-zZ=pQ;aJ@#<6n^EUC_r_HC^A?RDFz$}fD6Q9{nNDmh*i~8Y(N-}wSUgda+-E|
zZu%L!!`wea7h6w6iLYRfV{4m|Gi!J0%{!-3Q)va6Ki8-9S;juI(MPdN8VvV;z>0Ys
zlBBn+`?hh1WaKPZisoJe`jK-$8Lnm@(`>STG{n2lCHOkGJ_O4mGI#+I0^tJ>wt+JN
z@?(YYwrC5zeCvd&O)|~0u@USN6g#3x-#vi-|4IKX`_B~p7dUUs^~u8o!NJaSIr$eRq|VCl>%tR*8|>k|
zmwyU*Y+(iRs=YiaJTkKMk}2wN#fp)Vjsq?v5D*vEBZ~xgfF8*-qV^4ZbVCqDZGL54
zDXRq{0tJw31l%q-&snFJmlv5w#=C++r~TZzsi@TretD-b5dBc-o_D+(DT$e_bJ)M5
zBAQ-VX`$a{KfgR>(*!7TpCC5T)l>sK(G*WB{>d&nLCJcz{|Lru-cuQxAGeRM(Zja6
z4o%c&=i)&B=O8z1yHqN>wQmbQ^$|;eE}T=1=jPPa)ls#+kSP(Cj4wlC1`}+O!(=O1
zj0n%vLZ}BH$sFJZE7ryGZA!1|f;_JoO_7?%F8#8S@Q_{?0o0EKZ=u_x6FP-NZ#)AS
zvA1K*96EoQ{Q@HUq)YLUBMqco*9EaRhhHfv`>rpvNP^NSY(NjDld}8Qs$MsMTkPKD
zR32GSAjZlodu%Tqy~x>xiZ1cn@$r;T)@G!ot)PoN+oj>GiNj#_5^?I^;wy!{)~rQ}
z(gf5@nQu@}%f;h&c5VG8Ug~zYGy&G2`4&3Nmz-IU^s(uiuz-rEX)ai`Ip-;LJw*%H
zB(6K{*`uMBIS=+;3liKS8MS#hIXS2?^P3|LDvh>u*zSJQA(_Eh6zX|hB_5m
z=5sHwm{AL)%n2SwmRqG+G0hT@t0*b+=AG}%w?Mxm9{{Dw-~W2h(JgAj$z?h_EG-oj
zn0bI+{>WZE?lfk|WaRsKXW^fsX74NP;tiEq|I&1PHvY4~x!tr1R&JG7*{$7Nb6M1V
zNf5B`zV1Q!qd1^3mev}jko7btPwsy5MAT+(ZGCv)=&v#6u{neaZ|zAQm9A@M#pHH{{#
zPy>O7Ep%(NzqrNF`(}^P<`%FCVd3GIhxRex;^{)=yHGE(&quU=YS1nAR(!zKqL(jK
z{UsfW>gt&`dh;&i=DPouc&&6W2F-HWxyX8h;>%5g!^6Tx)^HISc~nQT=?8!t+qa)u
zT<5a6WrBA{>-52mLUsLiRafvJ+l?g^quD)oS#f0AEYbyzIORiq5Mnp&LU#b_(F&?*
zXh4%ByZjJDTJ!M8%)uft_xtbv@G_5GoRqOL^3@~S&V{Y*1=6DIbtj34Q2bVS;;sRm
zIKil{riNcG#d@6YWVSqjuZ;3JDMZ*r6?_u0YGaDQ>t?&Pv)4rCBevtT>+0(lk49K_
z9ZRP2ER+NRk`Wy9SZeal+nUEX-J;5zJ{_erpDN7n{la5A>dHNeoP)0q-_5?JXC50+
zLe?K{a&WzRtGarY78r|11m@J;XT+Yrkkj5hzhUu`Gp2((`GJkGkjYdiVdkI5zok$J
zPChW;tN@fi!fB*b3C!8KmPw@}_XAuxHr{5%r;}8m^i@M_kZYN0vCC;?lP$12NQ11f
zbe#Lpf9!sB_4OC_0#p-#Dd(oyh|4)G$(ncxZ`q?QkC$5SdT_O9>AkF@G~P@68gCRi
z6OxCR4<^^HiGOTo6ALB%01FMXiQfA~eJInJA`u=2%@*;}p$EjxD;q9Ys2ORo3Ntvk
z5m*Mtk1wN>KPLUN8MSDLht#Z?WMhuW3&asU_ErMR7M3LP0ZRNYpbO^DpWm^k)pe5l
zV@wHzQ;Oe<0(M?$L;KueO7|lS4CTchjUh_D919r>#n$bM4>Zp42Z_Rn06?VHd_C4*
z;@Xecr8d07jSQ~%)(}-<#jGk-7~j$-;bScQJ|DcHK3#xzwB&6DdxKpDFC50Az}r#}
zB@kvD7}~CZkBSoh>jhtq9oT1n4!6H{c_)
z4QeI9AmH4rF3SD&BRpdg62p=+gd)xpI*zwa*VY!iLwF|wVjC7}1Odnd9De|?f#)cJ
zugUy~Gk)?Py|R~+$s6iUqW(XR@7m*d2I2Mpcrf`_&Rn}y>WS;U)ei)06mW)WRvT>B
J%Pl-&{s)h1dnW(@

diff --git a/docs/articles/extras.html b/docs/articles/extras.html
index 5bcb8f6..b18e4bf 100644
--- a/docs/articles/extras.html
+++ b/docs/articles/extras.html
@@ -88,7 +88,7 @@
       
    Extra information about the package
    
                         
    Krystian Igras
    
             
-            
    2019-08-31
    
+            
    2019-09-05
    
       
       
       
@@ -110,24 +110,24 @@ 
    
 
  • “auto” compare increasing and decreasing approximation and chooses better one (basing on \(R^2\) statistic)
    • 
 
 
    Let’s see below example:
    
-
    library(randomForest)
-library(pdp)
-library(xspliner)
-data(boston)
-set.seed(123)
-boston_rf <- randomForest(cmedv ~ lstat + ptratio + age, data = boston)
-model_xs <- xspline(
-  cmedv ~
-    xs(lstat, transition = list(k = 6), effect = list(type = "pdp", grid.resolution = 100)) +
-    xs(ptratio, transition = list(k = 5), effect = list(type = "pdp", grid.resolution = 100)) +
-    age,
-  model = boston_rf,
-  xs_opts = list(transition = list(monotonic = "auto"))
-)
-
-plot(model_xs, "ptratio", plot_deriv = TRUE)
    
+
    library(randomForest)
+library(pdp)
+library(xspliner)
+data(boston)
+set.seed(123)
+boston_rf <- randomForest(cmedv ~ lstat + ptratio + age, data = boston)
+model_xs <- xspline(
+  cmedv ~
+    xs(lstat, transition = list(k = 6), effect = list(type = "pdp", grid.resolution = 100)) +
+    xs(ptratio, transition = list(k = 5), effect = list(type = "pdp", grid.resolution = 100)) +
+    age,
+  model = boston_rf,
+  xs_opts = list(transition = list(monotonic = "auto"))
+)
+
+plot(model_xs, "ptratio", plot_deriv = TRUE)
    
 
    
-
    plot(model_xs, "lstat", plot_deriv = TRUE)
    
+
    plot(model_xs, "lstat", plot_deriv = TRUE)
    
 
    
 
 
    
@@ -143,20 +143,20 @@ 
    
 compare_stat - function of lm class object. It defines statistic that should be used in decision between spline model and linear one. The function should have the attribute higher. When the attribute has "better" value then the model with higher statistic value is chosen.
 
 
    You can see the feature in above example:
    
-
    set.seed(123)
-boston_rf <- randomForest(cmedv ~ lstat + ptratio + age, data = boston)
-model_pdp_auto <- xspline(
-  cmedv ~
-    xs(lstat, transition = list(k = 6), effect = list(type = "pdp", grid.resolution = 60)) +
-    xs(ptratio, transition = list(k = 4), effect = list(type = "pdp", grid.resolution = 40)) +
-    age,
-  model = boston_rf,
-  xs_opts = list(transition = list(alter = "auto"))
-)
-
-# aic statistic is used by default
-
-summary(model_pdp_auto)
    
+
    set.seed(123)
+boston_rf <- randomForest(cmedv ~ lstat + ptratio + age, data = boston)
+model_pdp_auto <- xspline(
+  cmedv ~
+    xs(lstat, transition = list(k = 6), effect = list(type = "pdp", grid.resolution = 60)) +
+    xs(ptratio, transition = list(k = 4), effect = list(type = "pdp", grid.resolution = 40)) +
+    age,
+  model = boston_rf,
+  xs_opts = list(transition = list(alter = "auto"))
+)
+
+# aic statistic is used by default
+
+summary(model_pdp_auto)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = cmedv ~ xs(lstat) + ptratio + age, family = family, 
@@ -191,19 +191,19 @@ 
    
 
    Link parameters stores info about what function should be used to transform the response. The transformation is used in the final model fitting. The standard link is the identity (for gaussian distribution) - for binomial distribution logit is used.
    
 
    See more at ??stats::family.glm.
    
 
    xspline function allows you to decide which response should be used in the final model. Let’s check the example below in which poisson distribution with log link is used.
    
-
    library(xspliner)
-library(randomForest)
-x <- rnorm(100)
-z <- rnorm(100)
-y <- rpois(100, exp(1 + x + z))
-data <- data.frame(x, y, z)
-model_rf <- randomForest(y ~ x + z, data = data)
-model_xs_1 <- xspline(model_rf)
-model_xs_2 <- xspline(model_rf, family = poisson(), link = "log")
    
+
    library(xspliner)
+library(randomForest)
+x <- rnorm(100)
+z <- rnorm(100)
+y <- rpois(100, exp(1 + x + z))
+data <- data.frame(x, y, z)
+model_rf <- randomForest(y ~ x + z, data = data)
+model_xs_1 <- xspline(model_rf)
+model_xs_2 <- xspline(model_rf, family = poisson(), link = "log")
    
 
    Let’s compare two models by checking its AIC statistics:
    
-
+
    model_xs_1$aic
    
 
    ## [1] 672.5753
    
-
+
    model_xs_2$aic
    
 
    ## [1] 580.0274
    
 
    As we can see the second model is better.
    
 
    
@@ -211,15 +211,15 @@ 
    
 
    
 Transformed response
    
 
    In some cases you may want to transform model response with you own function. Let’s check the example below with random forest model:
    
-
    set.seed(123)
-x <- rnorm(100, 10)
-z <- rnorm(100, 10)
-y <- x * z * rnorm(100, 1, 0.1)
-data <- data.frame(x, z, y)
-model_rf <- randomForest(log(y) ~ x + z, data = data)
    
+
    set.seed(123)
+x <- rnorm(100, 10)
+z <- rnorm(100, 10)
+y <- x * z * rnorm(100, 1, 0.1)
+data <- data.frame(x, z, y)
+model_rf <- randomForest(log(y) ~ x + z, data = data)
    
 
    In this case log transformation for y, removes interaction of x and z. In xspliner same transformation is used by default:
    
-
    model_xs <- xspline(model_rf)
-summary(model_xs)
    
+
    model_xs <- xspline(model_rf)
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = log(y) ~ xs(x) + xs(z), family = family, 
@@ -244,7 +244,7 @@ 
    
 ## AIC: -193.88
 ## 
 ## Number of Fisher Scoring iterations: 2
    
-
    plot(model_xs, model = model_rf, data = data)
    
+
    plot(model_xs, model = model_rf, data = data)
    
 
    
 
 
    
@@ -253,14 +253,14 @@ 
    
 
    When interactions between predictors occurs black box models in fact deal much better that linear models. xspliner offers using formulas with variables interactions.
    
 
    You can do it in two possible forms.
    
 
    Lets start with creating data and building black box:
    
-
    x <- rnorm(100)
-z <- rnorm(100)
-y <- x + x * z + z + rnorm(100, 0, 0.1)
-data <- data.frame(x, y, z)
-model_rf <- randomForest(y ~ x + z, data = data)
    
+
    x <- rnorm(100)
+z <- rnorm(100)
+y <- x + x * z + z + rnorm(100, 0, 0.1)
+data <- data.frame(x, y, z)
+model_rf <- randomForest(y ~ x + z, data = data)
    
 
    The first option is specifying formula with * sign, as in standard linear models.
    
-
    model_xs <- xspline(y ~ x * z, model = model_rf)
-summary(model_xs)
    
+
    model_xs <- xspline(y ~ x * z, model = model_rf)
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = y ~ x * z, family = family, data = data)
@@ -285,11 +285,11 @@ 
    
 ## AIC: -153.1
 ## 
 ## Number of Fisher Scoring iterations: 2
    
-
    plot(model_xs, model = model_rf, data = data)
    
+
    plot(model_xs, model = model_rf, data = data)
    
 
    
 
    The second one is adding form parameter equal to “multiplicative” in case of passing just the model or dot formula.
    
-
    model_xs <- xspline(model_rf, form = "multiplicative")
-summary(model_xs)
    
+
    model_xs <- xspline(model_rf, form = "multiplicative")
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = y ~ xs(x) * xs(z), family = family, data = data)
@@ -314,10 +314,10 @@ 
    
 ## AIC: 94.46
 ## 
 ## Number of Fisher Scoring iterations: 2
    
-
    plot(model_xs, model = model_rf, data = data)
    
+
    plot(model_xs, model = model_rf, data = data)
    
 
    
-
    model_xs <- xspline(y ~ ., model = model_rf, form = "multiplicative")
-summary(model_xs)
    
+
    model_xs <- xspline(y ~ ., model = model_rf, form = "multiplicative")
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = y ~ xs(x) * xs(z), family = family, data = data)
@@ -342,23 +342,23 @@ 
    
 ## AIC: 94.46
 ## 
 ## Number of Fisher Scoring iterations: 2
    
-
    plot(model_xs, model = model_rf, data = data)
    
+
    plot(model_xs, model = model_rf, data = data)
    
 
    
 
    
 
    
 
    
 Subset formula
    
 
    Every example we saw before used to use the same variables in black box and xspliner model. In fact this is not obligatory. How can it be used? For example to build a simpler model based on truncated amount of predictors. Let’s see below example:
    
-
    library(randomForest)
-library(xspliner)
-data(airquality)
-air <- na.omit(airquality)
-model_rf <- randomForest(Ozone ~ ., data = air)
-varImpPlot(model_rf)
    
+
    library(randomForest)
+library(xspliner)
+data(airquality)
+air <- na.omit(airquality)
+model_rf <- randomForest(Ozone ~ ., data = air)
+varImpPlot(model_rf)
    
 
    
 
    As we can see Wind and Temp variables are of the highest importance. Let’s build xspliner basing on just the Two variables.
    
-
    model_xs <- xspline(Ozone ~ xs(Wind) + xs(Temp), model = model_rf)
-summary(model_xs)
    
+
    model_xs <- xspline(Ozone ~ xs(Wind) + xs(Temp), model = model_rf)
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = Ozone ~ xs(Wind) + xs(Temp), family = family, 
@@ -383,11 +383,11 @@ 
    
 ## AIC: 960.62
 ## 
 ## Number of Fisher Scoring iterations: 2
    
-
    plot(model_xs, model = model_rf, data = air)
    
+
    plot(model_xs, model = model_rf, data = air)
    
 
    
 
    Or model including variables interaction:
    
-
    model_xs <- xspline(Ozone ~ xs(Wind) * xs(Temp), model = model_rf)
-summary(model_xs)
    
+
    model_xs <- xspline(Ozone ~ xs(Wind) * xs(Temp), model = model_rf)
+summary(model_xs)
    
 
    ## 
 ## Call:
 ## stats::glm(formula = Ozone ~ xs(Wind) * xs(Temp), family = family, 
@@ -413,7 +413,7 @@ 
    
 ## AIC: 953.43
 ## 
 ## Number of Fisher Scoring iterations: 2
    
-
    plot(model_xs, model = model_rf, data = air)
    
+
    plot(model_xs, model = model_rf, data = air)
    
 
    
 
    
   
diff --git a/docs/articles/graphics.html b/docs/articles/graphics.html
index a666aeb..a1edcb8 100644
--- a/docs/articles/graphics.html
+++ b/docs/articles/graphics.html
@@ -88,7 +88,7 @@
       
    Graphics
    
                         
    Krystian Igras
    
             
-            
    2019-08-31
    
+            
    2019-09-05
    
       
       
       
@@ -104,9 +104,9 @@ 
    
 
    As you actually saw in the previous sections, the model transformation can be presented on simple graphics. In continuous predictor case, the transformation is plotted as estimated spline - in case of discrete predictor, factorMerger plot is used.
    
 
    For quantitative predictors some additional options are offered. Let’s check them on the below examples.
    
 
    Let’s use already known Boston Housing Data, and a random forest model.
    
-
    library(pdp)
-data(boston)
-str(boston)
    
+
    library(pdp)
+data(boston)
+str(boston)
    
 
    ## 'data.frame':    506 obs. of  16 variables:
 ##  $ lon    : num  -71 -71 -70.9 -70.9 -70.9 ...
 ##  $ lat    : num  42.3 42.3 42.3 42.3 42.3 ...
@@ -124,29 +124,29 @@ 
    
 ##  $ ptratio: num  15.3 17.8 17.8 18.7 18.7 18.7 15.2 15.2 15.2 15.2 ...
 ##  $ b      : num  397 397 393 395 397 ...
 ##  $ lstat  : num  4.98 9.14 4.03 2.94 5.33 ...
    
-
    data(boston)
-set.seed(123)
-
-library(randomForest)
-boston.rf <- randomForest(cmedv ~ lstat + ptratio + age, data = boston)
    
+
    data(boston)
+set.seed(123)
+
+library(randomForest)
+boston.rf <- randomForest(cmedv ~ lstat + ptratio + age, data = boston)
    
 
    Also let’s build xspliner model with lstat and ptratio transformation.
    
-
    library(xspliner)
-model_xs <- xspline(cmedv ~ xs(lstat) + xs(ptratio) + age, model = boston.rf)
    
+
    library(xspliner)
+model_xs <- xspline(cmedv ~ xs(lstat) + xs(ptratio) + age, model = boston.rf)
    
 
    In order to plot specified predictor transformation just use:
    
-
    plot(model_xs, "lstat")
    
+
    plot(model_xs, "lstat")
    
 
    
 
    You can also add some additional information for the plot.
    
 
    Plotting training data
    
 
    Just add two parameters: data = <training data> and flag plot_data = TRUE.
    
-
    plot(model_xs, "lstat", data = boston, plot_data = TRUE)
    
+
    plot(model_xs, "lstat", data = boston, plot_data = TRUE)
    
 
    
 
    Plotting approximation derivative
    
 
    The derivative is plotted on scale of the second axis. To display it just add: plot_deriv = TRUE.
    
-
    plot(model_xs, "lstat", plot_deriv = TRUE)
    
+
    plot(model_xs, "lstat", plot_deriv = TRUE)
    
 
    
 
    You can also specify if the model effect or transition should be displayed. Parameters responsible for that are plot_response and plot_approx respectively.
    
 
    For example below we display just the approximation and derivative for ptratio variable:
    
-
    plot(model_xs, "ptratio", plot_response = FALSE, plot_deriv = TRUE)
    
+
    plot(model_xs, "ptratio", plot_response = FALSE, plot_deriv = TRUE)
    
 
    
 
 
    
@@ -154,52 +154,52 @@ 
    
 Models comparison
    
 
    As we may want to compare the final GLM model with its parent black box, xspliner provides one simple tool.
    
 
    For comparison just add model parameter for plot method, and provide data:
    
-
    plot(model_xs, model = boston.rf, data = boston)
    
+
    plot(model_xs, model = boston.rf, data = boston)
    
 
    
 
    The resulting graphics compares predicted values for both GLM and black box model.
    
 
    For predicting values standard predict method is used: function(object, newdata) predict(object, newdata). So for regression models the results are on the same scale.
    
 
    The notable difference occurs in the classification models. GLM models by default return “link” function values, so for classification it can be any real number. Contrary to that, randomForest function returns predicted levels.
    
 
    To avoid the problem, predictor_funs parameter was added. This is the list of prediction functions for each model (in order: black box, xspliner). Let’s see it on SVM example:
    
-
    iris_data <- droplevels(iris[iris$Species != "setosa", ])
-
-library(e1071) 
-library(xspliner)
-model_svm <- svm(Species ~  Sepal.Length + Sepal.Width + Petal.Length + Petal.Width, 
-                 data = iris_data, probability = TRUE)
-
-model_xs <- xspline(Species ~  xs(Sepal.Length) + xs(Sepal.Width) + xs(Petal.Length) + xs(Petal.Width),
-                    model = model_svm)
    
+
    iris_data <- droplevels(iris[iris$Species != "setosa", ])
+
+library(e1071) 
+library(xspliner)
+model_svm <- svm(Species ~  Sepal.Length + Sepal.Width + Petal.Length + Petal.Width, 
+                 data = iris_data, probability = TRUE)
+
+model_xs <- xspline(Species ~  xs(Sepal.Length) + xs(Sepal.Width) + xs(Petal.Length) + xs(Petal.Width),
+                    model = model_svm)
    
 
    Now we specify predict functions to return probability of virginica response.
    
-
    prob_svm <- function(object, newdata) attr(predict(object, newdata = newdata, probability = TRUE), "probabilities")[, 2]
-prob_xs <- function(object, newdata) predict(object, newdata = newdata, type = "response")
    
+
    prob_svm <- function(object, newdata) attr(predict(object, newdata = newdata, probability = TRUE), "probabilities")[, 2]
+prob_xs <- function(object, newdata) predict(object, newdata = newdata, type = "response")
    
 
    And plot the result
    
-
    plot(model_xs, model = model_svm, data = iris_data,
-     prediction_funs = list(prob_xs, prob_svm)
-)  
    
+
    plot(model_xs, model = model_svm, data = iris_data,
+     prediction_funs = list(prob_xs, prob_svm)
+)  
    
 
    
 
    It is also possible to sort the values of heatmap according to chosen model:
    
-
    plot(model_xs, model = model_svm, data = iris_data,
-     prediction_funs = list(prob_xs, prob_svm),
-     sort_by = "svm"
-)  
    
+
    plot(model_xs, model = model_svm, data = iris_data,
+     prediction_funs = list(prob_xs, prob_svm),
+     sort_by = "svm"
+)  
    
 
    
 
    In case of class predictions, let’s create class prediction function first:
    
-
    class_svm <- function(object, newdata) predict(object, newdata = newdata)
-response_levels <- levels(iris_data$Species)
-class_xs <- function(object, newdata) {
-  probs <- predict(object, newdata = newdata, type = "response")
-  factor(ifelse(probs > 0.5, response_levels[2], response_levels[1]), levels = response_levels)
-}
    
+
    class_svm <- function(object, newdata) predict(object, newdata = newdata)
+response_levels <- levels(iris_data$Species)
+class_xs <- function(object, newdata) {
+  probs <- predict(object, newdata = newdata, type = "response")
+  factor(ifelse(probs > 0.5, response_levels[2], response_levels[1]), levels = response_levels)
+}
    
 
    And plot the result:
    
-
    plot(model_xs, model = model_svm, data = iris_data,
-     prediction_funs = list(class_xs, class_svm)
-)  
    
+
    plot(model_xs, model = model_svm, data = iris_data,
+     prediction_funs = list(class_xs, class_svm)
+)  
    
 
    
 
    Sorting values according to specified model is also possible:
    
-
    plot(model_xs, model = model_svm, data = iris_data,
-     prediction_funs = list(class_xs, class_svm),
-     sort_by = "svm"
-)  
    
+
    plot(model_xs, model = model_svm, data = iris_data,
+     prediction_funs = list(class_xs, class_svm),
+     sort_by = "svm"
+)  
    
 
    
 
    
 
    
@@ -207,23 +207,23 @@ 
    
 
    Following above approach it’s easy to generate similar graphics for higher amount of models.
    
 
    Just include additional models inside compare_with parameter (named list), and add corresponding predict functions to them to predictor_funs parameter (if omitted, the default one is used).
    
 
    See below example on airquality data
    
-
    library(mgcv)
-
-data(airquality)
-ozone <- subset(na.omit(airquality),
-                select = c("Ozone", "Solar.R", "Wind", "Temp"))
-set.seed(123)
-
-model_rf <- randomForest(Ozone ~ ., data = ozone)
-model_xs <- xspline(Ozone ~ xs(Solar.R) + xs(Wind) + xs(Temp), model_rf, data = ozone)
-model_glm <- glm(Ozone ~ ., data = ozone)
-model_gam <- mgcv::gam(Ozone ~ s(Solar.R) + s(Wind) + s(Temp), data = ozone)
-
-plot(model_xs, 
-     model = model_rf, 
-     data = ozone, 
-     compare_with = list(glm = model_glm, gam = model_gam),
-     sort_by = "xspliner")
    
+
    library(mgcv)
+
+data(airquality)
+ozone <- subset(na.omit(airquality),
+                select = c("Ozone", "Solar.R", "Wind", "Temp"))
+set.seed(123)
+
+model_rf <- randomForest(Ozone ~ ., data = ozone)
+model_xs <- xspline(Ozone ~ xs(Solar.R) + xs(Wind) + xs(Temp), model_rf, data = ozone)
+model_glm <- glm(Ozone ~ ., data = ozone)
+model_gam <- mgcv::gam(Ozone ~ s(Solar.R) + s(Wind) + s(Temp), data = ozone)
+
+plot(model_xs, 
+     model = model_rf, 
+     data = ozone, 
+     compare_with = list(glm = model_glm, gam = model_gam),
+     sort_by = "xspliner")
    
 
    
 
    
 
    
@@ -231,13 +231,13 @@ 
    
 
    
 Plotting more transitions at once
    
 
    If you want to display many transitions on one plot just pass xspliner model to plot:
    
-
    plot(model_xs)
    
+
    plot(model_xs)
    
 
    
 
    For models trained on top of many predictors, there will be displayed plots for 6 first transformed variables. To change that value just set n_plots variable:
    
-
    plot(model_xs, n_plots = 2)
    
+
    plot(model_xs, n_plots = 2)
    
 
    
 
    You can select interesting variables to plot just passing predictor names in vector:
    
-
    plot(model_xs, c("Wind", "Temp"))
    
+
    plot(model_xs, c("Wind", "Temp"))
    
 
    
 
   
diff --git a/docs/articles/methods.html b/docs/articles/methods.html
index ab29d78..77c892d 100644
--- a/docs/articles/methods.html
+++ b/docs/articles/methods.html
@@ -88,7 +88,7 @@
       
    Methods and xspliner environment
    
                         
    Krystian Igras
    
             
-            
    2019-08-31
    
+            
    2019-09-05
    
       
       
       
@@ -101,42 +101,42 @@ 
    2019-08-31
    
 
    
 Predict
    
 
    As xspliner final model is GLM, predict method is just wrapper of stats::predict.glm function. Let’s see it on the below example:
    
-
    library(xspliner)
-library(randomForest)
-library(magrittr)
-
-rf_iris <- randomForest(Petal.Width ~  Sepal.Length + Petal.Length + Species, data = iris)
-model_xs <- xspline(Petal.Width ~ 
-  Sepal.Length + 
-  xs(Petal.Length, effect = list(grid.resolution = 100), transition = list(bs = "cr")) + 
-  xf(Species, transition = list(stat = "loglikelihood", value = -300)),
-  model = rf_iris)
-newdata <- data.frame(
-  Sepal.Length = 10, 
-  Petal.Length = 2, 
-  Species = factor("virginica", levels = levels(iris$Species)))
-predict(model_xs, newdata = newdata)
    
+
    library(xspliner)
+library(randomForest)
+library(magrittr)
+
+rf_iris <- randomForest(Petal.Width ~  Sepal.Length + Petal.Length + Species, data = iris)
+model_xs <- xspline(Petal.Width ~ 
+  Sepal.Length + 
+  xs(Petal.Length, effect = list(grid.resolution = 100), transition = list(bs = "cr")) + 
+  xf(Species, transition = list(stat = "loglikelihood", value = -300)),
+  model = rf_iris)
+newdata <- data.frame(
+  Sepal.Length = 10, 
+  Petal.Length = 2, 
+  Species = factor("virginica", levels = levels(iris$Species)))
+predict(model_xs, newdata = newdata)
    
 
    ##          1 
-## -0.3811033
    
+## -0.3923137
    
 
 
    
 
    
 Print
    
 
    Print method works similarly to the summary. In case of passing just the model, standard print.glm is used.
    
-
    print(model_xs)
    
+
    print(model_xs)
    
 
    ## 
 ## Call:  stats::glm(formula = Petal.Width ~ Sepal.Length + xs(Petal.Length) + 
 ##     xf(Species), family = family, data = data)
 ## 
 ## Coefficients:
 ##                    (Intercept)                    Sepal.Length  
-##                       -1.96540                        -0.01363  
+##                      -2.244882                       -0.009104  
 ##               xs(Petal.Length)  xf(Species)versicolorvirginica  
-##                        3.13004                        -0.62584  
+##                       3.340926                       -0.654606  
 ## 
 ## Degrees of Freedom: 149 Total (i.e. Null);  146 Residual
 ## Null Deviance:       86.57 
-## Residual Deviance: 5.14  AIC: -70.36
    
+## Residual Deviance: 5.172     AIC: -69.44

    Predictor based print

    @@ -144,7 +144,7 @@

    Standard usage print(xspliner_object, variable_name)

    Quantitative variable

    When predictor is the quantitative variable its transition is based on GAM model. For this case print uses standard print.gam method.

    -
    print(model_xs, "Petal.Length")
    +
    print(model_xs, "Petal.Length")
    ## 
 ## Family: gaussian 
 ## Link function: identity 
@@ -153,21 +153,21 @@ 
    
 ## yhat ~ s(Petal.Length, bs = "cr")
 ## 
 ## Estimated degrees of freedom:
-## 8.82  total = 9.82 
+## 8.83  total = 9.83 
 ## 
-## GCV score: 0.001429038
    +## GCV score: 0.001315672

    Qualitative variable

    In case of qualitative predictor, standard print.factorMerger method is used.

    -
    print(model_xs, "Species")
    +
    print(model_xs, "Species")
    ## Family: gaussian Factor Merger.
 ## 
 ## Factor levels were merged in the following order:
 ## 
 ##      groupA       groupB                     model   pvalVsFull   pvalVsPrevious
 ## ---  -----------  --------------------  ----------  -----------  ---------------
-## 0                                        -269.6600            1                1
-## 1    versicolor   virginica              -283.9595            0                0
-## 2    setosa       versicolorvirginica    -354.5249            0                0
    +## 0 -257.6859 1 1 +## 1 versicolor virginica -273.8312 0 0 +## 2 setosa versicolorvirginica -352.9511 0 0

    @@ -184,31 +184,31 @@

    Extracting effect

    Each transition is built on top of the black box response data. For example the default response for quantitative variables is PDP - for qualitative ones ICE.

    In order to extract the effect use transition method with type parameter equals to data

    -
    transition(model_xs, predictor = "Petal.Length", type = "data") %>% 
-  head
    +
    transition(model_xs, predictor = "Petal.Length", type = "data") %>% 
+  head
    ##   Petal.Length      yhat
-## 1     1.000000 0.7207022
-## 2     1.059596 0.7206924
-## 3     1.119192 0.7206924
-## 4     1.178788 0.7230503
-## 5     1.238384 0.7248216
-## 6     1.297980 0.7266156
    -
    transition(model_xs, predictor = "Species", type = "data") %>% 
-  head
    +## 1 1.000000 0.7501526 +## 2 1.059596 0.7499859 +## 3 1.119192 0.7499859 +## 4 1.178788 0.7531005 +## 5 1.238384 0.7549714 +## 6 1.297980 0.7583468 +
    transition(model_xs, predictor = "Species", type = "data") %>% 
+  head
    ##      Species      yhat yhat.id
-## 1     setosa 0.2440234       1
-## 2 versicolor 0.6806043       1
-## 3  virginica 0.8353717       1
-## 4     setosa 0.2169328       2
-## 5 versicolor 0.6781563       2
-## 6  virginica 0.8346654       2
    +## 1 setosa 0.2437544 1 +## 2 versicolor 0.6966708 1 +## 3 virginica 0.8499297 1 +## 4 setosa 0.2157778 2 +## 5 versicolor 0.6923198 2 +## 6 virginica 0.8465953 2

    Extracting transition model

    After we built transition basing on continuity of variable specific model is created. In case of quantitative predictor we build GAM model in order to get spline approximation of effect. In case of qualitative predictor we build factorMerger object and get optimal factor division on that.

    To extract the model, use transition method with type = "base":

    -
    transition(model_xs, predictor = "Petal.Length", type = "base")
    +
    transition(model_xs, predictor = "Petal.Length", type = "base")
    ## 
 ## Family: gaussian 
 ## Link function: identity 
@@ -217,31 +217,31 @@ 
    
 ## yhat ~ s(Petal.Length, bs = "cr")
 ## 
 ## Estimated degrees of freedom:
-## 8.82  total = 9.82 
+## 8.83  total = 9.83 
 ## 
-## GCV score: 0.001429038
    -
    transition(model_xs, predictor = "Species", type = "base")
    +## GCV score: 0.001315672 +
    transition(model_xs, predictor = "Species", type = "base")
    ## Family: gaussian Factor Merger.
 ## 
 ## Factor levels were merged in the following order:
 ## 
 ##      groupA       groupB                     model   pvalVsFull   pvalVsPrevious
 ## ---  -----------  --------------------  ----------  -----------  ---------------
-## 0                                        -269.6600            1                1
-## 1    versicolor   virginica              -283.9595            0                0
-## 2    setosa       versicolorvirginica    -354.5249            0                0
    +## 0 -257.6859 1 1 +## 1 versicolor virginica -273.8312 0 0 +## 2 setosa versicolorvirginica -352.9511 0 0

    Extracting transition function

    The final result of building transition is transformation function, that is used in the final GLM model estimation.

    To extract the function just use transition method with type = "function".

    -
    petal_length_xs <- transition(model_xs, predictor = "Petal.Length", type = "function")
-x <- seq(1, 7, length.out = 50)
-plot(x, petal_length_xs(x))
    +
    petal_length_xs <- transition(model_xs, predictor = "Petal.Length", type = "function")
+x <- seq(1, 7, length.out = 50)
+plot(x, petal_length_xs(x))

    -
    species_xf <- transition(model_xs, predictor = "Species", type = "function")
-species_xf(c("setosa", "versicolor", "virginica"))
    +
    species_xf <- transition(model_xs, predictor = "Species", type = "function")
+species_xf(c("setosa", "versicolor", "virginica"))
    ## [1] setosa              versicolorvirginica versicolorvirginica
 ## Levels: setosa versicolorvirginica
    @@ -254,7 +254,7 @@

    GLM summary

    Standard summary method is just wrapper for summary::glm. In order to use this just type:

    -
    summary(model_xs)
    +
    summary(model_xs)
    ## 
 ## Call:
 ## stats::glm(formula = Petal.Width ~ Sepal.Length + xs(Petal.Length) + 
@@ -262,22 +262,22 @@ 
    
 ## 
 ## Deviance Residuals: 
 ##      Min        1Q    Median        3Q       Max  
-## -0.68602  -0.08184  -0.01356   0.10264   0.50215  
+## -0.68163  -0.07833  -0.02037   0.10712   0.50143  
 ## 
 ## Coefficients:
-##                                Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)                    -1.96540    0.14013 -14.025  < 2e-16 ***
-## Sepal.Length                   -0.01363    0.03631  -0.375    0.708    
-## xs(Petal.Length)                3.13004    0.21589  14.498  < 2e-16 ***
-## xf(Species)versicolorvirginica -0.62584    0.12133  -5.158 7.99e-07 ***
+##                                 Estimate Std. Error t value Pr(>|t|)    
+## (Intercept)                    -2.244882   0.146470 -15.327  < 2e-16 ***
+## Sepal.Length                   -0.009104   0.036244  -0.251    0.802    
+## xs(Petal.Length)                3.340926   0.231635  14.423  < 2e-16 ***
+## xf(Species)versicolorvirginica -0.654606   0.123765  -5.289 4.41e-07 ***
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## (Dispersion parameter for gaussian family taken to be 0.03520518)
+## (Dispersion parameter for gaussian family taken to be 0.03542157)
 ## 
-##     Null deviance: 86.57  on 149  degrees of freedom
-## Residual deviance:  5.14  on 146  degrees of freedom
-## AIC: -70.357
+##     Null deviance: 86.5699  on 149  degrees of freedom
+## Residual deviance:  5.1715  on 146  degrees of freedom
+## AIC: -69.438
 ## 
 ## Number of Fisher Scoring iterations: 2
    @@ -288,7 +288,7 @@

    Standard usage summary(xspliner_object, variable_name)

    Quantitative variable

    When predictor is quantitative variable its transition is based on GAM model. For this case summary displays summary of that model.

    -
    summary(model_xs, "Petal.Length")
    +
    summary(model_xs, "Petal.Length")
    ## 
 ## Family: gaussian 
 ## Link function: identity 
@@ -298,21 +298,21 @@ 
    
 ## 
 ## Parametric coefficients:
 ##             Estimate Std. Error t value Pr(>|t|)    
-## (Intercept)  1.20102    0.00359   334.6   <2e-16 ***
+## (Intercept) 1.207191   0.003444   350.5   <2e-16 ***
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
 ## Approximate significance of smooth terms:
 ##                   edf Ref.df     F p-value    
-## s(Petal.Length) 8.818   8.99 756.3  <2e-16 ***
+## s(Petal.Length) 8.827  8.991 733.4  <2e-16 ***
 ## ---
 ## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 ## 
-## R-sq.(adj) =  0.986   Deviance explained = 98.7%
-## GCV = 0.001429  Scale est. = 0.0012887  n = 100
    +## R-sq.(adj) = 0.985 Deviance explained = 98.7% +## GCV = 0.0013157 Scale est. = 0.0011864 n = 100

    Qualitative variable

    In case of qualitative predictor, the method displays data.frame storing information how factors were merged during the transition.

    -
    summary(model_xs, "Species")
    +
    summary(model_xs, "Species")
    ##         orig                pred
 ## 1     setosa              setosa
 ## 2 versicolor versicolorvirginica
@@ -324,15 +324,15 @@ 
    
 
    Providing model parameter instead of predictor, the summary displays a few statistics that compares original model with surrogate one. All statistics definitions are included in summary.xspline documentation.
    
 
    Here we show one example for classification model.
    
 
    For this example we use ISLR::Default data and build svm model as black box. The model aims to predict default variable, indicating whether the customer defaulted on their debt.
    
-
    library(xspliner)
-library(e1071)
-set.seed(1)
-data <- ISLR::Default
-default.svm <- svm(default ~ ., data = data, probability = TRUE)
-default.xs <- xspline(default ~ student + xs(balance) + xs(income), model = default.svm)
    
+
    library(xspliner)
+library(e1071)
+set.seed(1)
+data <- ISLR::Default
+default.svm <- svm(default ~ ., data = data, probability = TRUE)
+default.xs <- xspline(default ~ student + xs(balance) + xs(income), model = default.svm)
    
 
    In order to check the summary, we need to specify prediction functions for each model. In this case predictions are probabilities of success:
    
-
    prob_svm <- function(object, newdata) attr(predict(object, newdata = newdata, probability = TRUE), "probabilities")[, 2]
-prob_xs <- function(object, newdata) predict(object, newdata = newdata, type = "response")
    
+
    prob_svm <- function(object, newdata) attr(predict(object, newdata = newdata, probability = TRUE), "probabilities")[, 2]
+prob_xs <- function(object, newdata) predict(object, newdata = newdata, type = "response")
    
 
    Almost each summary statistic compares models basing on some data.
    
 
    In this case we’re going to compare models on training data providing:
    
 
      
@@ -343,32 +343,22 @@ 
      
 
    • 
 prediction_funs as a list of prediction functions (for surrogate and original model respectively)
      • 
 
    
-
    summary(default.xs, model = default.svm, newdata = data, prediction_funs = list(prob_xs, prob_svm))
    
+
    summary(default.xs, model = default.svm, newdata = data, prediction_funs = list(prob_xs, prob_svm))
    
 
    ## Models comparison 
 ##   1 - Max prediction normed-diff:  0.5268109 
 ##   R^2:  0.9185403
    
 
    ## Setting levels: control = No, case = Yes
 ## Setting levels: control = No, case = Yes
    
-
    ## Warning in coords.roc(roc_surrogate, x = thresholds, input = "threshold", :
-## An upcoming version of pROC will set the 'transpose' argument to FALSE
-## by default. Set transpose = TRUE explicitly to keep the current behavior,
-## or transpose = FALSE to adopt the new one and silence this warning. Type
-## help(coords_transpose) for additional information.
    
-
    ## Warning in coords.roc(roc_original, x = thresholds, input = "threshold", :
-## An upcoming version of pROC will set the 'transpose' argument to FALSE
-## by default. Set transpose = TRUE explicitly to keep the current behavior,
-## or transpose = FALSE to adopt the new one and silence this warning. Type
-## help(coords_transpose) for additional information.
    
 
    ##   1 - Max ROC diff:  0.8712113 
 ##   1 - Mean ROC diff:  0.9506292
    
 
    Another set of statistics is generated for prediction functions that return response levels.
    
-
    response_svm <- function(object, newdata) predict(object, newdata = newdata)
-response_xs <- function(object, newdata) {
-  y_levels <- levels(newdata[[environment(object)$response]])
-  factor(y_levels[(predict.glm(object, newdata = newdata, type = "link") > 0) + 1], levels = y_levels)
-}
    
+
    response_svm <- function(object, newdata) predict(object, newdata = newdata)
+response_xs <- function(object, newdata) {
+  y_levels <- levels(newdata[[environment(object)$response]])
+  factor(y_levels[(predict.glm(object, newdata = newdata, type = "link") > 0) + 1], levels = y_levels)
+}
    
 
    And similarly to previous example:
    
-
    summary(default.xs, model = default.svm, newdata = data, prediction_funs = list(response_xs, response_svm))
    
+
    summary(default.xs, model = default.svm, newdata = data, prediction_funs = list(response_xs, response_svm))
    
 
    ## Models comparison 
 ##   Mean predictions similarity:  0.9966 
 ##   ACC Black Box:  0.9719 
diff --git a/docs/articles/methods_files/figure-html/unnamed-chunk-7-1.png b/docs/articles/methods_files/figure-html/unnamed-chunk-7-1.png
index 55bc1ffd2ad4620eda2dca62271364676e17ec8a..7d49c8d9e115cf5f1fd4c6e898e920b29cac928c 100644
GIT binary patch
literal 21606
zcmeFZXHb)2_cw?L(nM4cl&aD}Kt-g3h;->qT2N_HLnm|v0i{Y8=^a9o5<&?jAfQy~
zJp?JCC-fTf-s1D@e?RQZ&g|}&oq0bvljOe2xv%S#-}#+`pS3iUs3{pKiHL}(Rg_=o
z5D}3=h=?w~A-@d#VmHI?0DK|Rda0)Xe7$t(5-~CH<;#~zNJvOYNv~YFLPkbLPEJli
zK|x7LNkv6<_3Bk>YU*p(uF=rYT)%$(#*G^{Z{DP(rKO{zyLIapJv}`G1HRH{?%jL$?%ltCpPikZgM)*Ulaq^!i<_I9hlhukmzR%^
zkDs6a!Gi|^0s?}9f!Hm2?+^F
zNl7UwDQRhG85tQ_S=ndLp2^9{J%9dOUS9siix&zC3W|z~N=iz~%E~G#DypigYHDgP
zU%pgVSJ%+c(A3n_($dn_*4EL{(bd(})6>(}*EcXQFf=rL_3D+8k&&^nv5AR^si~=%
znHdNKGB-DW{rdHrH*em)eQRN1VQFb;Wo2b;ZEa&?^X}a{TU%Q@J3D)Odj|&xM@L5|
zCnqo%?Ck99;^N}!>gwj^=I-w9;o;%w>FMR=1%W`ky}f;We0+U<{rvp={rv+10s;dA
z-@kwV;lqcZprGL3;Ex|ahJ=KKhK7cPg?;+;>GS8$;o;#C5fPD*kx@}m(b3T{F)?4h
ze2I;Xjf;zmkB?7CNJvafOiD^hPEP*%^=nE>%C~RdQd3jY($dn?(=#$MGBY!?va+(X
zvvYECa&vR@^78WY^9u?J3JVL1ii(Phi=j|xNl8g*X=zzmS$TPRMMcH;@82scE32xi
zs;jGOYHDCGSZ!@B91gFmtE;cCZ)j*}Y-~gz5KT=@&CSg%EiJ9BtwRsU0pwa{`~dpS9f=JPft&8Z*N~;Uw?o9z`(%o-@gY32Zx4+hKGkoMn*dAeCMG5)C#R;Srl+T8W@ct*XHh8B+}s=*jh>&M|MTb1!otGh;^NZM((>{$
z27_5ySy^3OU0YjQUtizY*x20M+}hg0VzD?JZhL!sXJ==3cXw}ZZ-0OP;Nal!@bKv9
z==k{fQBO+o6`}^-w%+j_M5z$>Dl^1e)
zkhG0iuh(=Pb6wj5`C%xMJl>$7_muauaz2HJ+|S|h>>$qJQP85}rQc|>z*18koK(M7
z(0Y-eq)n0l9gxwe15@rz{C9|H@BLu
zAhBpU9PvZTVQI1g6_1kvo~I4x1^$=i3>pc&ag9QR76p^YQ3GfPZCHlKuj-C-oe9@U
z&1SCC4G#SPE(tMC-DIb0waf5v)H*!ybr8cd@vnz_0$wfpeuI;?jjF4Xh7@PMXKjdy
zl|o6i$tZgg#V`pA&*NKt#4e_ri>ub%`xW@D_7u*spJr3C(vahgH)HJ
zuo_>sXlR^li+k85OKtN@YGs*_RCniPq#xcks%OBnFr~(ocvd*zPBErQCAqnGP7RW-
zK_s1PkJf)O?+bp~Iy1ie*?wbEYsXA)qx7)-!)9uY_^kSZz!@Ry@g%iU_c+
zb+++HXLC@WlV^GQ@d}~1;4qhssaXDbl(2j1!n@6VkZlW32XNOnH;k=JnXV!;Y4oFh
zQX4R=th@^fNYB{!E%Ip4h*X?SsF?}L9*YrgcqAp(6kQ&%@vy9R%}jDX#)$S%`y{w>
zwb`B}dLa}n7XHc|A@sy&5nPEpQ}b5^yS-X}a;&IO!YidxKTbaz`F*46Ds@!k!QhQp
zs{1{uO6RI|=F)G;OiXmM<-JBkE42yNXY2dSQ^f8Fm6n}X+2
zR@}F_)1K~QO%I)^%Y!j}Ge-J?kmF&)uu+}9ft*|n-+Ffo@wAnpTN0q*%IP1``Hp#;PD<`E%rP-sGS%J^+Bvmo4JzDqb^BE`S9-Z!LGtFW53`3
zW+4p`J$)>5jc_!ixK{gM!g$gU)>b%v-lZpgvrU&z>9M<`V-st-L(aF5Xh{%*1fQgG
z1qnR;$><}WbVV|F`hKy6d)C%2m#C-2<2L#qn13M|P
zUwkfybXmJ)=jf1SldA+Vrj|DklCJ2l{IzrVn;Y5b6019h`_M%S?+Eq^D!=L!_7N@@
zq8_oSD*+gBidn%-SdcdSkEvC_O-5f?KNRC*!>HituqUmLPGZ(b;9i~O#aDXA%aNP=
z(Yn?xfk_9HPMM?H(SxaN$&}|feGolMuYLJ%xKqSXHKvQJq}&ME$z;dmi!OC$Yr5lb
zH8M?`EQ(TT19{T>$fre++)4BILiC`n$fG*^OPEFAs8peus??|fB_NBDe{t!GY?V^+
zug=#Cyy^23R@GHO)9{FnZ3a80ndZf_s%}EYHA*LT!bh5v{HMrGk@@wudzl}>ypU&j
z=G^gj&X7Hlt`*V-uDYfdu)1I^cqQ#X&gQ71I2$Jt>9lrB
zyd^_dP4fsIfhN^5P!eaQ=XuB0Tq|Bb5fZcnWtl1O?>1#-Vqv`2)H3XL{y8dxNoLwl
zwM#M@_f{-TXjnaZ5DWwJ=3pUgb+
zsmq{oGSnNNj?R3sNW6cRJ|f~C+|Cd4I6KgUP=JVBOrGrfDruY;T3VJx?}NxQhtKmu
zqgT95PQw_vS9@2`72>w@WFLL)wcOpUt(PmZZT<9_)%uv(&UJlFD%X(}p$kX%XBTd#
z=LjM^cP5dSCWX5MF?FqlDUmz4+D}
ze_opt(K9$K!7>gSOFWI{4h00-Hu6$;Q)#{ygHci}5Nk=IOq({K5&72bBI}6#=~&mY
z6w2O*HJ{MIxA$dM
z`J!?c-PO(@TmDIUa$Tt5>CJ|$?dXHZ^T;#8m?4DoGy7Yw+ZEuF*m>*uU)_x2Lf`Qx
zw~tw)SsW(cN22>!BxP5MT&4kA-u^oE3_4rvXjQzBgZs_jAxLplP1jJi4%(z|V%kv9
zkB|~Md;Tq2Kw`j*@Y)O(l|e5vuRE7h(QGhhq)Vol0mI*uJb5q?+0Y;{!Z*7);`_%X+
z%Ugxg<1ajawmcEIUQuSJJj!+Jbl>QFr0(80UZ`K+ACT1Lh|iEL;HFsW^wCVY-Oc70
zG&{@Mghe`E7^8Rk$Wz4AKf>aJiI!*Qzee-8%ntezqdjN92XAi`Z_D{wsrXy)E6Qx?
zVr@T=m`XLr`6%`)W$M9xQd*ZxDQ;;@4=E2Q$UbEkkaWN2ZB;IG@Devux~C<6T*j{X
z?kJ)eX-_MITW{|D$@xjrKgC?Z^Q(@Uh@js&zDu)bXsaJ1Z=J!aY$npwTX{9ao;1Gb${hee9fLxd!N$KuuLTL5#O=K5u!Eoa6WdYBg21pVDpC0Qw}`hW<)ARC<7l8
z7zPBsTCA$aXAn8Q5VUyTiCf$i+Q{|dTZtxBMZ6ZK@=qA=kw-V2zg7Z*1ZNHyo4wt@
z#;Ox%R?HmQcTE%s=bsdsNn|C~fcI1u+J^Oxw{n~X{fa+GI9`o8Z>;d7@!t!poswou
z6IJn-={dw`t_sXDOX7k*RB!dBKh87mFXSTTh8w2)Ev~}*=AdX6AVZlbZ&N(s8*9|<
z3So@bUmyR8&V{ai8cE5O5Oo|~KaX=+>yE=>90eck>6Bt>Hba8nZ{;Op6L+rKk5|^b2f0Ki+F8gYzt!y4s?nWx
z`;L79jfJFXiLT>_{s-We!|N4N;G}FgQDs#f+`l=Y9|Wh$cn5H|(a1&gR;q`^$@GGmS8ZLQVBa2gbT<+&4sn
z7~Z!p+clF`q6(*kht52Ncyr(QXS#1(lcv8U3&iBFnXty(_?HY%xDm*E#CzJx-_kTBX&O0E@NqG>o$h>nQ2=CU_NYLd^OtYYO^k_
zZ@%7Z5}_~M0e5DL<=QWIN{nG8I=6|{2u6eO?Ypt$pt~N9jLbf|Umpv7Zl=ZXn{U-9
zpTcU=Nb>`4N&+vUGj`J7EkVfbh>#JJln+zDf^L_717BeZiCJ7Ru*fMFAV1)>llN>`FsqO7)4$q$!UX`~XW=L6
zed)8I#+D^
zRCTT)sq&zM5|10kULoZg`mF_cLxK1E(E#U05enR>cN1Yb*km5&?2`xb)_=!-B(+(l
zlctUJ2)B?<^BVS0#TkT>cJc4hs)gip0DF;w4Rw8zcgZ~Zw3d=Baw`n9JMRVe%kr3w
z&Q%(kP~L}ppu@QoXdLAhRUSRbmwfDrDxOrpAjR>7nK5o{R&r3*9}^QzcJ{g4D8sT_
z(QL-MV|W@w-;!DT-$$ug!MX4ade(e?hcRM6EAoP#p4=TfR&AUrA(0~y(wZ;Rx9TDm
z3-PG=0TSSOjw8MW`n=?PFzQvyCr6H?(u1hgR982KEl)s6?)@FUWVkO7xu0ZXS>0A|
z1^lqRKYa8GL%;vsfc*L2{Z3W&10riMdgEbF?LW-*Ag5ahYFn}|aQ8*hb3=!pm$7+{
zs(9LeNDek^n8AmUgYtsTY%35)9bA~71x~71!g)>yf$2+e|GML72mr&2UK>O~b!*O_
z!*&@Yuj~r;Q8racTl_3=47gbU;e`+kBLRf6zIL5Nh!3$!ouAZoXur;1`DX&VaP8eh
z*6_P64z(*FN>F0==sj|fo_jTPEejmM*m9uVNj5#r{HcZg388gI!dbYB_zDdksx
z#bN2SG~2pgu42eE>ec~-4}uR7EhqsUtv?B3m{1K+Y_QPYZM$+Z3lf-k3MQ?XBN$!I
zH@jszl!7L$tfC_DRPIu%mtu;ZHq!pCM9H
z7q1B%GQge6j!goINIgsy$A(})1Q5t1dGS-kz-Cj$j122i+4xOh61oGuFC_CyA`aN)
zAhDvGpfJ_LxMq|Jdz~tE2XQWt>jx;8PFk^ew0e#?f`IY=#JKV{*pm9wnz7RM~Na~gR5Vk^WW4EWk#Fc?7;7Q<^w3ee*
zwJNOFsakktBz&1`;bc&-7WVKo#h&fGh-$Q76cQ6fbrtmW#Hzb1-f{e
z2ueK$8jXl>;|JAd)H66wT=PL@&vU`_n#5d>>iV_!t|BizTCT#*9SVafH|Q@L^10#;
zcpZ;HPCQDa+qyoUdsAC;>U(t^I%orhYp6p_iORfh65XaA&%7D)?p5~hPIZAeo%^Pd
zjjn4W#pN>JE20Gob%@#-*TY_MSA^0}@sm1vOG3=pA==uD$B)xv@;wIW?DJbif*5dl
zBGBax)-o?ged0rMP-`lmZGP^!j&PYi@%Z1b(ZAy>`jx&vNy`DvO#YA7a_}pT@c3MlIbED{FeLUu&>`{#nsDYZSCfO#A%+5e?|M1m>O#kS
znTUZ5pN*!+h?_*7-nibP?cMX=H9u{mm7+&@r;rHg^uK`h#dvlWHFoHHxy?!i3bYSm;KevgGpv^a
z$#VUs**rCf3iZg$>)oh!5Cf~8EpB?fcuL?b~_d5;~?`Rmb
z1H_SEi@C<#tJ3EUyG&{f7;vB$gcgt>_D%D%gTjK~{PqP0lhPaH}D6m{vB
z7ETbdHzEza<*#pIbr{c-7;;X+Y7P_!(^$fZyUPIQAeU?8woQ
zgE(N!j_!LW>nIuxN~=q#$9`Vfzu4xliKMDX7}g(=7Yg8nDeh_+XZ>t{o>K#
z$4b~&E<#{ZeH5}W0Manxy3rUtzFX!K=x5LC_%&+1i9{&Qn4yVroo-}!;6PQ*n-7Eb
zw_zA2_TEW%V_>R{t5);|2>~@HCUCy6Nly_51
zF}jV7GMVQ?t9R^LW;`G92pI&y?;>?OW6%m!sm!1qsog)+wQ6$pk8UFH<84|JtUK8^9}d@7QLZ_lC^SS#m*^m>7?3
zhD#4V1S)Vsx{q!klvzWBSijI84Sye2Py;M-IFbZQE0UjtZ1mnO?@wTC&iJ>9ZRrY^
z{+hG0ZgIC-t^J2L*3G6SK4}S$1P5d-GFyLPvqbVY-jE|BEGwPBqk%OmY(cUIiCClQ
z(bfmLULr4{{4lnZd$8{fwZgy{mr!E?N>Q-ER)vB62is03Zj%6G9DT}Q>P@M1NkHNs
zhqm=@x#`uT18;dB4!puSrCb-A)sIM_$mrt?Vo>>WlRCm9$=bTC){wGS_!dHVO0|BL
zxDa6>ntZYldL5yBb8{~w*Q*FsWkB)ZAuS?RkxavjCT$@wJ88tuE{NMb7_AUjRMNuS
z^r7GIJnN6}O@w)Ef~ym+j}}CQhIcbAkXFj-47kY8SBy7S(%YX~7I!G)3RKeX^GnE@
z>f0s#I}(+RibIyB=#$^ZoZnk}Ow}M0^U-^W84*Yr&mbHJF9B~e^ph!$OCKeU%e0=GEGG?3bi*02
zw0N4FmfbNI`KNO?Q@l0Ic$;?p4cPdCsDbI}`j}3u{7u+K=hRvnI%VM*t}W?IXFzUD
z=nc!Bo?buF1QIBce+X&*K}*X7T8S;UJ7nO3EQ?)7e1E(3d)XH1J|XX2OC{&Zb0{=U
zpZV>cl_&Gg8BK)f37(08fIIsVoubpp8=*#k9N&k$kwezCf{*v0ao+;%8|gVnFMElc
zi9-!mV8u{V@tGp;$xMn25e*WdUVbUFI~a9-T!l|<6rcawMn;ay?C^N}>6v&V5A>J#
zfRRb*2Xu(HXo=5}V}&=SI65(K?cT53jWLIC4EDCim{IN)i2H8ptL
zt$UC+rkJ-8X@_IGV|;Q$fbNYS1w~}a(lummw2}<05CbwPt_Quc@mPAY#T}f2#*3Rl
zSte?AVB>{e$s1)KlyGfzldF@J5s2g!<%s;dMz<+YZh2&s`185E2>dy1qWNml`;Xb^SxvL>r-y37#HqrgycdcT;=Mjyd`_E697Xy;_lvKCVX)VF@XGt6-hNg_nSXb{gr|TqrD72LTPMBaw-6mH;}X|}
zllY#EtH=9Me#EAFRii;{y9ax;gkE!NNGay;qP5kO`1nbWi*hQYN#}#dA8mI<|ex*d+am=4_`k;%dOr@2|?;LIJ@;~9BU%4;ik^Vrt?nIc)d1shI2KU@5;GyAR>I<#c|js9pl%L+{bU)u#@69%vuArn<`
zI9)pNgVlTaW;@^ug-ZQ#e*5~7uaPa;%Wwh9M#N{IP=*^#qV0#!{5Dlx2((JEm&&2=
zAT~xRcplJP>NH086Y-gVI_SuQ3yV?*X)Y=IGB*Trt`@D}E95=2_J&5x>}s<2G^d=y
zpi0eBjnm2EvvvYFR$Jh*FO5Z!oO%IqpnN&9rH+X8LK=Yg87p4lXI=QEI7?K*e
zm3=9DY+d4KWlylgqoEmBrF66?-gBM)zr6rXDGW|{0~K~w1NbW&M|FJdrrC$GI~1t1
zn|K2M$-qbWO(_>Ca+Kw$E`!3fEI_WFLz~UoSsHskl|N`+RxaKwlB|$;EZ^As*f6(U
zrBZ3+Od&9Q3$8;d)q{GzV&$9V+K#Efe}6QF_b6y&3Z_7D>>K)Qg}=f+H@h(dDtQ#`
zA27O=t$3~tC1`x#SF;g$ki+pIE2y%%-9#wC90r%k8a~M@H!$T3;jxp7To21QRAS6R52=
z-VuOq!Utu;hxZ3DA>0hUrGFUBuG^Nge7iL>K8_uDk#qpGi%4&NV8BXonfZ2HLs)&)
zY&M%3OXTA|iMh@v100|R;~m@no*@+llOTB85s#9ixiR*T;H+Js!BboCXT}B`=n%ee
zFAYcztY1hnrLHtN?+!F21Z+shv&vLgt#RV!&I3n=Z%_r78I_HWRX)JMx%H{@{Mxe-
z?}*C*(P`Xg*2SNn{M5wx7|gSo6Y3g{e%BOya2ri`&5P9LWb=41{}>jrJog$#BV=70
zzy_lz+xVW>==m-4yH3pL;JmI20V*vrbu6EX32dqRo)AMcJ)e)3R=YE3@^pMNXSswN
z6mw!@ncG_b)@S@zhob1LuhyVs5gC5&K&H7M|9}Rc8rUAv9d(v*yUZY#X8@ZI`V8GC
z(~s~kemZOT>raWXwhM4~+$!qqgwnkhb`LXkoX?QSX#UA5{8y4sg4nvinPhn)Ti~>M
z{blo%lm4%Y-BAyQ8f)W(YQ+|I7*7P{I--R%GnMyqqsP^vQ6dzQ
zf;!*+5vhY=uh(y1vxG!kZoiqmecLpS$P
zs}ana45xVWwboZQ#|!QKw|AbImV6axb~tRs7LQpZD=9?}^GCC_)E>d@pG`veA(xjC
z$0{{VOCd-l?AF>aNrlI{>ky0Aig%XhF-^OMLx#{A5~!?8@h`46f(o=p#m
zPuya&D;p*%+vorEB`FgPkYc0nLCSp1(^;B?V46YEjb-+Tz#47i7?aF~g`)?XJG#7?
zZj3^6D(1J2c4b$2%@@y(ef49~&jT`(svGuVh@Gw@92AXOcONTz#T5&bPo4a9r;k;KRK#IjGmx$9;EGfFLTjUCz`1>bitt$7z)LUSE4=AU;h8r5Z&S#-H#
z+Qc^`({w#sVpl$x-*qV9(^lm}(amw^-LB4V>D@|ER!L#QsQZ$LTxpA#UZ5~-JHI^c
zp}&;Y1xlR|mG|w})PpHIh6Z^}Mi@rBE035~;4x;@=4KDS=Si$(6xC}#Whyxe_`Kh*
zhBhM(tkw3(^feDyEbWF$ibx#*iSmYvr={F4t0)iUcleMHvr#d>eo4J?u>+PId-gur
zlTZWh9_C(B5JzYV+CM)uTxGQ_rG2AT%C?M1Zj0l}_5g>y(q
z+YRr9>-Tr&S9gDzv#fYWpe$EzUf;%AWaQHi)>Gu}X@>tgxbnlf&&+l^+s8SB%DV!-
za_+d8v>I=uBDiU50)CYI{@HV2x$~in_y7~S_@a6PjdTf*Q(tlCC{YT(xZDGJh`W7aQqt;kD=)}_-06#HqNv!2wY*~?P`I5P1G7cw5G
zv82*HooUY0ga(8+29_sun5W#Stk_A?*I%g(qZRp8IbgX@;WR!z+B1kG8*zwWAd;#}yG3cxGjpK{fiQYREExhK0e7ddNT8j7u?Nsos3FnKGs+bAkNL-9aEH
z?aK2;g`w3&_`j{bKSqvKu|iLk-r)lFJvN4fst_9UtKD}_HM1&@;;vGQ7;)-E#iEhv_x4Gn27?(a5B0oA=@pY$x2FNIre
z*32@9^%;nHfkf8>WjpuJ?`3TO91T;*
zD6hmq5#y?e>IERXvA4d}z2b)1^*Z>awJQ!l+%;YdT3|DK_OWbzZK>X0M)AJoQb=|e
zudtA*Z6jS?mg_ZCu&QAvI~MYZ7V+uT3A%&Z{0u#`i6!SqOHK+YRmTw*k^hI5BypV1
zjd1e-NdNc#ia}mTVvHlgp8f&vdgZvsNVK%p3mozH0cD(OzZXQdG|5Wls064`4OCx8
z92WVc&}1a5;D~VzJJ(R$;mYGuH#~M(C9fmS$4Dx=95T*3!`$F4OkY&HB_V1}H?q|x
zV`K`}6Igho(IOiIyHEx^8`uRL_h9T22PqTMWTEFGT^TP`Nz8lNS;aCI@K=jw37Z!SX6u`F6
z9s~4Kdnjj5L?s0%zW)O$bYK0Qf1r9PE+Q6frf=^1fE=D&
zOAR_cEsJa5g8YG8M{q?MhDF65jQ5K|FZ1>*<3MG6m9p+s(DlUDWiBhn8J~e2MclU<
zPf?a?!I86q48dxXt6yFKv>D2^nL7vGKzc4If6`KR_y9~&p)EUR5KoqLwA<3iqb3cX
zD8Hi22$*Adx$|D9U14kBxQaRF1*Kx$H>;`R)R5ppQBU;(a5r`meS?zniNpl6{yShk
z9?14rMpdxHL>s~Wt~c=5)2>QgnQiIg#>XAe&}fO?!v&
zvrU*YpD^jW0Q(@KgOvL8(hT>BLgX?=No6$oq(xGrLGt2YNMC1WlkxDWL5B*aLHfkg
zy`m8Xqg#Z3$3e6qTq7}gc={sqwvz;>*yg1`ZV0Lls042a!yZ^7O$xW-xJFm;dN}*q
z!)8rfsS8jwn6Pn`Mg@(InTx0adL9$<`%>$}AO@nV!gL6=3|8;h_A`=!JAX^Q{?yV)s~+AKL6k9GcezzFdzuSBmB?s;ME2kkhrP~kvJf1gYEyPS!e97g`AM2
z#}^L)uQJ*se5LnQCi-)C5E?j%|1;KG43*SF-<>I+?QQki$;JyC?
zDhi@dfK2_;1_rlS_67$|30}q#1u=ZyRxCj;`I=+R0cw>)+6~#bv+zq(;4hK(G?^Ub
zsw4K?PHLsr{#0K(aomC7GKVCP+Q@a)w66YsNobdzQ^xs~Uq>uimR4aYo4fb0!ioDS
zn|ALxYy+dCv-WP7_myKs>x&^f^yV*($D1Jb;Y;rbwKQl}4U!Ge3wN$3x
zDZ1__)foeDA!-a{fgE%Q3rlPW9SOMp=#?2u6LmLW(`X87RDD4#l8u5CYt^lOalJhq
zZ_P*5D^X;2HWhTORDjV7v#^}SYT*v_hX3vE=ROi#*G1!
zTE5i@&)*#?nl@a=M$_#>K{w1{!xKj3vOvT1_W_d%FG3kSd{uaE=@L2W~?GJ
z0}||-!j>S-i_sxc&af>57Q?W+5U-KfENqV83HcbK2c)^3E}bw#>@Ne`>JBHk
z1utmN>(h)HOo{$QCR^!qO
zm56EHUG+lwA8ffOSzc+#on`reo|TKxpLMbM<*?~+Avg(^LN#;#Ffb=HB!Ixy=V6`_
z<1cT$&5eFs;pI&f=U3~!l=0W|8XUF;?(*3OycV7|v@%MekpXKpz1kRK
zIcl69phZZoJ=)yS%#&v4b}Z;KCaM6?9n8aaaM`)Zux7Qm#0!lh70zt3J8HeuUg2Y{
z_~BF&n8QbS1~aF#)0j45m=w5z<6%dpmXw4)#jKa-^pSWdU-$*DiJ@72$BkZUdl{y$
z+3dRxfA2OH?a12jVWZ&pEo7lS{{=}m>%t!$Vfu#c*X8zSDS&q7{$Ct0Gckjn
zCC<$~Z)}owE7iPwQ)JwpP+xSiunE^GJ_gc(rbN
zUzcXO_Fq-MRRjoc-mBDf0H(S?i-NDvh{1rqAkKfxJ<(|-NyVaH9^nn_<$putJqT8N
z|G!abi3Ooa7hCRqe{ESB2$j5Lh+ue)iR~!o#Yq7%EDBw=ZS}97(AxYzc+kot$j1G{
z-L=DB!X_Hp~l+L(Q7+T}aglW)pALZ8T}x
z&?nY=0=Tr}=IH5P#wKSuq3EdT$NB>i`{x=xyx$F}w2*)Ww)KAyp4F}xFy7hMRk{*)
z_@2dkVF}=477og$Bpdf(BL9}HCPKhLQfSM!{A
z%Ee!vPM<2$6*&Vahu5cda9Wr99~cxo)6#dGoBVSK_NlD*JRkNHc}Dt{1Ja@~ZR=Fy
z7Pi-NA;fr;QC}Y8=3(*u&BoWKzL!5Gr;zcKYY(erpJENo)60<;xNJNz3SOJEIE{Ui
zv`BjMmol#ZEa1z?Mh!pq_9Q%Xqb!N+U=pv3=yDgM9v5b?c3x5Wrh*t;S?W>>X6W=ZTDc%G%qXt_Ba-bp{Ip|0fo4p_!`i;q_
zq>sLFl;{>v(<&
zeRGsJk0gkpdbE3T)~sa;N}>#OIs|A_*?5WKol0)rRk)6Lm7K9qSNB~)*lnTqko0d0
z%o*AG=U13}EKU_S{|b+jEi*HCF%>BKJL0MRC7nx1sz5nz4-l)}`KXE+<)*=9JwIU1
zxq|u#`uy46xxc2rh?#XGV^$kGh#W-6>sRkvn=rgCG95xx
z44lNP`JsH%S<*&9z$Jpj66dBKy?4j;X#`aVGcSMYUu
zXax7i7t1#GqS>Vuc!@-tV6P4`O+@3^exZ&Ggw1ejG&P*j8*aU#F+EDe+iiht#PvsX
z`O#0%gLFA@6-ygPd(!}!(O@qXD``gCmc^cvS!;o3YdZ!w8U5mwwMEj9^9cEIUf6)E
z7n$-tnk5_P+}U+62QEZ!xftEV&J2e)^N((co=kM>uz4u~LuU2oen`0K`9?J9yr)~v
zN8?Mr4GE^XYcefut|-!VTe
z1Q}~PbXlHXkd@ak*Na7#pi3KD1@!+@=2Hp}1!0
zn@CjLtA=beKVl8NGxQy&WiI0G@kx&cRd2KkR+!R#JZ$|}G0i}fsNqDs&7kL(k;k%P
zRpsSA^%5_jWEF%fOH%X=mGdEd4Q;NEo7
zHibDF66Zb8B*j(*2gg?SGQ$
zRP%foOOQ@Lz(8&Qnge_TiW2o6=kY5LygJ(4>$;$5wAuu0qG^(&u!CNGW
zEPzhP3etXzssG)OphJaBY&crI-(>687zv>Bz=%yhdp2>aNbf_NafWO0DWu-(PFg$*aIINW`jTju=!%*FB!+o>XEjDV$Ygvh{$
z#rY1jnT*#m+$||-HJ6tbac=Euia+O;_Du&;-2^lYo$ld*`IbFR3-x4_?AhU
zt5UZR=3{)S(`$Va`zxqoZ`=>~me*Fa)}vs!^Yi@Q<}aFu$6k|O=Z8m~x&6Yl^B3>S4$k-FH-UY2(;I1$
zG^Tdg*vMW3ntv#&3M8AFVO)|sj^=Uij(&$i3Wy#
zM9<=;HTl!<5h6Te-inUuKMRis;*#9Cg%hn98*tdGHiV-fX-(2$559|s!7f^@?EdRdg%2t
zWLJ7X|D5S*HtF}nh$pmx!z{M#u19Vc;h{hjQ=7!L1YxjpO$yK3#=4irc
ztTpsTy~i=XemdLi{KF`6Sc5p&@++pKmN!Dp-eK#bH5GM4Z^-*Y+r7?
zo+}xrFTFuTbbasd{|lhOGtQ>p^J*-k`1~`Wo=Eut(6Jqe%JQk4D3io_!83(o&;n<;
z9CnF4=2Q8~%96{03X3DuX{aq`6LPPGOHhB!!FbPV%_P?7$ZLAl-t)_&Q0DVHeSO_W
z5>@TXm)#YLVD^obb!UFjkFWHTzt~JA<8gDf1iI^^&eUBhs@mkFC8d2m
z_bdum6SNj6(-*$V!GdGX=&g^_XudEP8p=npL052@5sW=#YML9Z1J~Z)$pHFptL2fY
z1m)`2KWcqvA0641Gn(%sYsSe8wHF8xS5ak9woqfzT)m}ntzCFz?}S7WLy2$EBUOgZ
zYqetR{O|(kNS*~jm<97E2E>lG@+Jb
zVnyr0+s07vl@d-&yDj(nF}k)mjsH6QleF0Op@ITjmIr?u_2(Mo64K-woGDa@T{KEM
z>iag42&8xgG>tfgt}Noz;UTBC&hBB+jP!QWXulx~Bd4jV*SQfWmzRSk1Eg?C`_C48
ze$v-x3=-W&VZ!^C&XTR(v|U#>7wc@_wx80>N>6%AGQM9_z$N_|bocOa8tjzKm{3ZI
zzaN|oUCnWJv-Xf?wDT|OWXfSX<&~Cn&(6{mdE$zp&9Rw>Ewk(`B9%|**(rvU=5=|S8q#Eb(X*tVoZ_~X+_WIA
zV8rW%w^43>IOnG8Gg~Gb_d@wa?$tSFn=|LOw(lgvydLf<378c-gLE)DkI^`I8n8g{8PqXeSl(qycw%nvCILsGbJk2iP{
z5=G+_^P4|<&}kkehKYi=6k~GRu>VW0;a;7z
zlwZ1tWllehlkciX*UC>^v*`Nb1iIlHTt(#zXz
zQ=)ifYu5bq*I}APG`zt^Z3>e_9Y6$9z&S1D;S#65ggv=iC#?t;Qv?s=(4J|#f~Ex2
z{5mYqx=?B0Ek4)7xy~cqE<}iCk;ZVJbbb)Qu4nr)tNM6TqdVLTYgW~wxBdkrkl#Is
zJ$(MOKefrOQI>*8B@#WW3KnPk0x>Qmh0C`a+R1AS`Xe`En!79QE5T5w6@F2ImEDv7
zRnC<^HI*lEm=RPEluK3-)Nv6(fnkTh$f&F+IE;cIM;u~69S#YEaQPB$6F}vNECQl(
zIY1BtBn%`Wn1o|gbb=7%3K9|mC<-AEjvSnshyTEC?fx{iRj=OB?|r+fyI*yE`}nlk
z35B;`{D-kooL)X87C3M`V6M7$s4l2TT~Qg&nEVmn5_ut4=}}Z}G>q60crE5h`9Vxm
zbjxh?5^4J=Rf+|&PP)xQ4Z3zx-TT{fQYUGzBQu>{G}?#{TK+DIRd-g^yzp9ta+=A_7A!*^=oDvqzA2FBUzR0)T*^tdSayh4
zIeYq_{z_|q%OI)&*)H1k03(dzx`LY`6f~Aqgbh_ePUbOv0LBkFbS7Op~L}E
zvB?0T(axd6{=^W=RT^x@tRFB+QfUKo{kG|OvhAhAfDq48(ajO^Z#R=2S2{o!tgb@Z
zOfvr`0pVq~8cNI`_kLZNJ7u${C6;GOe=yky$A(Rx&5uHiM
zgXV+VugPF=f$WcOcm{O4kjG+;E3j4H7Er$l%KacoHgN$w)5iHuUp`k;;ruVr=YLa2
zs0XErqf*}jS`qrp5fqo*%qW@1ik-&goU~McJuLqGE`?|HvhLZVl``?(ak~pPUPwbd
zGJYpIET6Yh$w6{;o_6|NNHKzV=L?hEN;L$|tLHYz`5x>DRe>nR1
z&6##yUuyg6i_$S1XXH8~AFN4kZzSQLf~~)-%Iw)gw~fd1WPV|c$MnP!Q!y||$JTw-
z>V;~8XB(wEdrp=J`EVaumf{^?GyjN@N9kNXlu+e*D4eTXm>|mvwSb3I+}6-x!u;p{
z4u)LPOMO$$cn&)E_IuJ{D{R1+PSMvq9E`vgg+nKP$AgjHWYFicJP6%D!Q~k&W5sf6
zB^hZracKkV(8CUmSKB7AK{nu&*CiejyZZu7YE^Lcf(zr-%b-wU{|U$b87sHnVA)1x
zLDf}vaOius-GG)A%c4Xd!4eiU-hEi(y}0wJ9`u$_ytl=BqC}{*Hia+@MdvalB`>%3
z4n-~oPifas>GRO9ZyHa6ed%8Dn+&A)zW@o!>;(vOQqm(5cjoOCRUDe#sfuLjGqH~F
zPcFl$L827aQ4S?k&wNpR${%Fn(c>n4aBs;&BCtwAG1Sq6L18E!B!A#>)!yr&(;B;&
zr_UvrV;qlQPB(P9*^nSJZdO{Gm9#?<+7Js)3X+*J%5(MTI^G8iTua?cV?OTI#>p(@&}
zBVWXvE1@J@<(YB-$UOa0d?i#dmkBH)p%Xvtq&CQ)YVBctg_f1;N-ErW4cE?_@kEUr
zo*Rv>pEg{t0?o4_%Inboh#Y%q(c4Aq%y_-)Sw`Yk?R~wLUc<;{ar701M?7!%!xr+%
zMfdG%HX_O}u*d3zT=4=;D~yku7=LqwkaC401^L7*f8?YU=@Fi*`DSHsnz5}}Xk|k+
zqy`_06-SRl>vNtD6_p67LF;dhnSXN0DL(12L-wYw3bh=ItRMSTc=O1}nUz66CyM@0
zwzlJS`7c!sTwYyAdQ+vVEZ7WV8HYABU!Pf~bskA|EAZw4T2Yr_=g*~e|BVI0A-xEO
zNbasX=LyNVGSB;H-=t}hLy@@4UE=47Nfz$Eu|7FubtoCtBD+T6doHwT1Z8EC)2@e@
z=NsF&#E*cy==|S?+D&`)?82gcA#tXWkz2?H*pxEJV*k%W0yaKbm04yxdfhx~Ay8uV
za5aIFMC+PM9+$V+I`o=MzXM~?NvqDMATzoA#SW`iFmlyO{_0feiJ+
z2R`>BwEV*{w9F+EXuo+6$>czu3(^ANMzkC?dr*_X(GUG3&)U1+K7Y0e?#8sDtQFye
z)T{OV(aMJbNUmNE%v2n;@_8dcjg}g2IC1X|j_df|3^MDuQk(Q4w@CJt`WD1VNqN>z
z?R)`G6K%3pOx#g}ZqZ@x-UnX`D|u2n;-m(tQ%X$iSXwG&-bTx*d_c2}q
z2dxqTppjhwE1oN`9$afrX@euEw?OBsBHZ)u2Uq>yOmQp6vFDyp$WC3a2^&C`YUV)@
z5Zl%5+v(UT)izx9jfk^$e;htiZWO6je!l0dp1VO%dq;NJ
zarPYj7AHnFBmHSq;W{UoIZkv4W2tWF`Nsn=$m)9)O=ry1P^Yh(FM^~f)`*CG@8_SxMSUH=Y`*W0tpEfY-3($XnHB~TRDVx1Jn&b!%xsOaw}PLlPdg0`
POH^@nc0WsUx{CV~a5dCC

literal 21587
zcmeFZXHb(}_cuxv=@yED6a_>;z(%i%NRuKExB+P*ARTE5gd!+#E7C!fD$)}|qy|VR
z5drDFLkQACN$4e%bD_`sod2AecjkOKAI{ACftlpm`&zrLz1DB7-wN*y^fZ_mxELrX
zD44Z09~)9o&_XCEsBOJv}`G0|O%?BNG!7Gcz*_3kxeN>xBy!E?&HN>C&akmoKxiv9YtWb8v8Qa&lg|
za^>pPt6W@M+}zyPu3h8d;o;@w<>TYy=jXqE{rZg?Hv|L(1O)|!goJM1yeTX!eCyUN
z5fKqlQBg55vD>$Ai;Ig(NJvOZO5VA1=kDFR_wL=hfB*i22M?sAq@<;#Wn^SzWo6~$
zXJ_~FC>l_l$6h(KY#i1B{elQEiElQJv}2MBQrBI
zD=X{k*RR>x+26i>%gM>f&CSir%gfKtFDNJ|EG#T4DuTgaa5%iUxEO&zl$4Ypk;w1g
zzn7MlmX(#2mzP&mR8&@0R#jD@P^jwa>YAFG+S*z)8jZnV>gww1>+2gD8X6lLu~_Vn
zA3vI!ntuNL+1%XR($dn}+WPC)ueP?f_V)IUj*iaG&aSSm?(XiMo}S*`-oC!R{{H@f
zfq}un!J(m{;o;$tk&)5S(Xp|y@$vDAiHXU{$*HL+91b@wJS3CJM@L7;$HylpC#R>UzrR}s0e67jQ`5|wf`TXV?2j^be#4%E
zf}cX`@k3)s`sx&Ix-p2aWy7&*G*98NDh(|?{Y5r5rtgjJw9Sw1oNs>gXtf2h_C&pz
zgYEKTw$cli*^}dp9|sgo`4=p*$dz|y-Ul!29S2$wYaEHVtq@=TYUtvm_dG4d6`=`Q
z3f27U5X#0oVVo390{`d3|Jj89pJ;;4GCo6LPL$&N+M$$IcU2M#PK=gfw=QYU6=>nf
z5Qxeskd{UH*%At1@PG~k0vXSy1<`mq{{vr;=}f_
z@w0*x?pMd7kK9JiDOiINqvHkg14I!MGey8sBaVVi{6pfTLR@po!OkPDAs$l-jCQdn
zyM5fmAjk6he!~YRm>6A_>Q~>&iSTjJiIw=2btQQV=qdZUqN-9;h}vzJZ}C_Y$*KAM
z`k?J~mR7(%2wHD&f-RMg-hXEOiqTJwFBrNTX`Hxg&89?>k7Zy{*&Ce79ros-YZw*u
ztTdZRWiJ)BQgS;=YE}S^$M&d@j$%tKtt;v4IDa=T`Fxj?^4$+pT_G&1xT;j}oB?>y@JLxWD;!0m;gM2r6ik|jlVEy#Ts;9cqI=3_-qbsDz`k3%S_DBu!ZXUF+t6=foh~U^?_VoOfQ*u{%!Pb*q<8&8uQgI*5c9nC`oFkjrCj*}
zOCM*q<u-(0DF@h0
zoH{=Olv+z#lQx#-h4+k>C|S2~>mN%tIU6f~{iT$xnA*Gj=&zW?4x1HP&}Ux4^8vHg
zNY-kTv6KTN$l6=bkiA+pxcON<<7iX~->9B5d!1S7_Y6fkmJrXmY%}LWXtnojHtcRo
zNRh?%b6;P^X{Yt#tsiu3JH18N;X9Uu{g6R*JL(DP)b*u0wYSt0Co_yM8h%HlR_%m{
zv(QMs$tWY`L3vZZ5BJ^GvR9^;iXy*4+SDJX8Qlz^T}GQ=1ixn>v}1^vM{mQXD*ihm=2&v@Ea$0_Oy}%iowt{tyIvRK7GM1K(ACu2jO&8+wUlHwV
zZ;Zamn-IvbemNiM(1$~e&)l)>d$JgrI`XQA#5_B+_M@&tOn#mIsc{~D(QhQ;xAKzD
z6yGRsq5icxR+x3+C-}8GH>vBROEBIUMv^t0;6!A(h@sG4Uw2Jb2vD6qE;A-?8H{`@
zE{kzQ@ebV$T@xV8?pNi(I7V@K9oOo%DqL*H2HW6}EY?A}@5g$+!x~Wi{=!B-%N@_d
zgQLt89t`4papP8HxeSDzjA8Lh-{X&_ch&FHiNQhDP*-AxTw-wX*{nm#-c}RAxuV%U
zTMi?=56h_#rtsx=vJQfmfyt}}`@pBxcT$>MEA3)#$AyezikPFj7I4|
zp{v_byG*Vc&XA`biH*DUzvhdlukdh=)+4H7aOFN0Hss=3xqzPY$2ARt4)tpp*%Le(
z1Rf3IfV{$Y9vEc|?rqb|B&an~vL$7egR65ft+rjkYpyOuCM70%_PuQwr|SFT^jUhk
zUo8Ve_pWH=40XKzb#DG#)nz-XO&(t+*J%))260RN-uF6%g_KjtM>6~+EYnHM`}YQ+
z84%fh>?Z@i`h4>(6vMDOyak?
z0zw2{CV4NAZ%yg(DWhv^_zSrutBk9oEUYTGI7mhEv{jtVPjQEj>CRO-E??Ck((qgy
z&E)YWO9i@-8$OdeGq$zAzs`-Jx1K0@#d~n8N1-ymeJ79ZV
z3=;8y<9i4M-oVE7#k!$;@@|lI5iI+)c9yC4tzQ6tvT?B3cV8Z?K%9St3(WV6k(6W#
zvRRnYk$eng?aRQnH2C^zzaosz!Q#izr^B&|ZFg>}+t(
z-$*{{rA3ZFV_2dm|EbgDdGsDC^>4J4T=h7w~%S~70I+wZiW7lxyMCSRR
zF@eMTI(cQykq?YjE7ds`qy@YwixNl-a@Xug^GmOz4Nk>)>a`5GEH%>nw90gMSyC@8
za0It0OKjv&0vij>Kj7U)j}$?UzC?En$&q&Ry1QwW9=)FYwNW_A?eDy|sZETC)!%r>
zJG^679Th+`;AI)Y{*KSUssdhnZUpj=-#6HkPXzzGkGlb{7Rs{tTw(WOga;Trin!7f
zNvi35F>ANOI2&W?xWDAD?8NLOF?JVjN;8VvUg*7l$QAJ;)E_(Qy1%d>Y<_vfM=Yy{
z&UZX#&B<<9fqZjh3mHC(c8~yvRBXYyV18XczhvS14Gxfx0~!!Ho1Arr%iM2aun)6W
zEA-ReQPnLEe(saTfO~qO!XBGkk4RLbrEk+ieDs)AWXYeo9_JhD{%X(SRC>A2)9SYr
z*xjKJ%&GQb7zefq^-8n6yftyU>YjbaZS6S+_=G+i*W@V2irb&0*M&Whd;Kw6XTpsi
zLb=-8FlPU4ixOir_+WH&PzxHVJRSgl)QKrEta#C(;Z?vS``xM4yVcF_?(y-zb_zT}
zT`3MQZ6++EO=NzvqHUFku!0Y^I`>OE%Ms*+h`c7Akrv>gt78WEPo4nNNChxlRB^*s
zy$7Ep2AbR;yI*(+EMu}vIKWS?VlS$1-m2cjr8Q;5Vg5-7{M3(BkA5J(-TfK-)s7C(
z$kGoUNH)v|#AUN_8{dbYW||n7N~iPtLicA`*PSGHSizZv9RhT0eE5Ij
zuL%Jo{M%@C~?)y?4ZX&)c!Y$>{!F92himC;5Rwt<1BoVo*Z&c`N=zMDTLh`jlj#y<5fN
z?VafLYxx&jblPcXB;&BW2=?@%?;T5;Bgh&v$;b&XjG$LHaB)JFM0FLur0CS}Y_@f}
zCe}ym<+Ijj?KMIqTx2qJ@=j_Uq5mqD@n!LKQNBPs&yD-EwIcmW{W-PA>osUFXkN)3
zUhx1xieCui;z_NUU4^ll2K9QRG#;;IE
zF61pcpdn)1K+F3ZR0ZyUA)Yqt2|z-bq%4@z(NYyJx#?~!14~lVHRWD(H+t=Yy%NUx
z%V4V_$_`om;gkE8f#BJ+3U#G+ynTo+L$f=pEoUPUa=gDO6ugS7)4s<7w8K<7AXW^h
zEW;XFX_vM;z(!nkuo|i|=^B0;;^{M0kUl}hN{8hBk|%t7C^;S&?{YC(Pqg`W{bv
zPgQ>?>!R!N5uv=fEUzVP_$`zJ01@Wc|gMKb@t
zc-nrPX!6vBDKS`n_^xv(AC@t~+1Doo#VGo=pW)RalPX*IRA-O+;d-VHJUnehZ<
z4cygXZykd%UR-C!;p(jQj?&IsBA7l>Ev!J7tMkaUT_Fo;*Bc5IRT@Yn1$hYIHpq(=Gn0W=dElQ!`l}>jL=T9W2}h6=vynbjrtr|
zMpq}+AdV@sI8aHimB^?xiWUjhGF|R%whH4MO@;fs@NxX^{ejx>-<7Ztiw8G!Dupj%
z8i#L-A`6<D1v>CK
zfRW
zg6Y!lb-Z}GO2`%~M^!9v;8zOFPv1?X$(JwGz^fG?)1&KUaRbQlM`Fw}raT6zdcB29
z-HR{_%$i!fbH5w}>gslY3+e>XCFQq!oz$9It60Ek=yewH#|OjWgCJJ#AUgoEPY@dIZDoB=ptu}HD
zJmki`T4x*|*7&Kt#6~0e{FQs~FnVTpntA(kmw6nUJA)Mx41z8vXUI`v898gU>%J*S
zx=2Ek`LzHw#*F|HGZzj8Ys-sjgHfgo2~GA`^z{_HdP`f%mOs8p3TrN;c`1b!NhAh`
zab9)E5pF#&&q~}bt2B%gbZ5k6)9K>AcnCvg9t~CYmZFgP|G;UqeplhAw1BA;Mta|9
z`nDQ2w5$El3$Vs8s9rRqfCA?-TT~vu9AM#w+?)cp$LU#P0-R<(9eWA@#!I}xaya*+a|TI!BZf8H;tLaQD5SGYi83Cq{$augRFdmD
zeL8+>i{?n;Qo|}NzQ4a)ywsop3XUD%&B~a&cv+QnqtZ>@N-iiLvDIDEkO|vbz*pGc
zfTZ9s8mr^;(H@7wZEstm@aoQJlOYwxDwAD<9FSks!y=Jm{PD(7;3@oDY9?eR4!vGq
zMbzq2te-mC3fA2GzHuEA-+$8C5p(;e!Kj|V?N4F-#^YFYfI-!u&*(e*zLHZMci;1N
zu2-|clGvyuzvuuBu(DCt4QSri2gzQf`T^A;i36{WxAn``mdrS<&0nwi0fFW&33eX(
zZu31X{SQUxkfKx7(To#PDC4R-8z1rN+VC5##0XsCPC})&0ZdyCf;6u}G+}V|rb7O6
zvKKI#PWF{8r6W2y2vC?|Gc}tk;nnTMf^MFj?6dyGG{`!ACcxfSc>4zVI6m}K-fHVl
zipQ_j-Ei0|VA0~e7fd5*IQ|%9d1*{eqrl=eZC;xu33l>xskJ5gRyV3R3o8i3eS!#}&Xeb*b-PlKI$~
zOsM}dcYRiabfdY;O<+iVvs#*$UD5IE-0t(0h|62GU#-U~)Fl&tqlh06*KE&Yql{4}
zxF@0Wx!sv==u1hwM&(>B!%4eW(*>js1lwkX-r>U;j}CWTj2OXzS0+0~eaiA24^O#z
zc*-Q&nuH?o>I?_OlcuymCuo~v)ykXKEn(o<==u&zsW0(!eU~1SZulyiiC5?m`gHsP
z@AQ8t!e`*jMVAQZIi4!zKjPKftIg%N!>u{>cnr06{KpW&ixG6&@xLIBVQMQ#NYS?diJt*
zw4I~2%|ZN5wqmmDS5?9|MZSeD$nFL#Uv)`tq!PfQ4?Lc!
z2>(aqP{m*=IG*>r3H@!G>pqPU4u`{~BbG>WI1NXV7wM#~P5?!2?g!3RJ-vWCM*-de
zx~{G`;bE?)IaYUCd2r6xY^&t
zDDt-tOja{8lJkeZaO+mlkhzVy2%X2qTlVCjLB=$tyQRC$Ic1;?>~F!?Yjk0pMK9dN
z*T|I(r3(?)m{FF@ouWIeBOK}R0`pbG(;r~Qz>o$`iT(BPlQr+W^Z~@=zzs6v3fQYeR?!KVFDvvPXpN
zW#p{vpO{Tfg`~lY@`_#p_wXb1InTEN9~D!3xChT;=R21%BT%ZSEETS!H`#Ux8c!*(
zjI%KVmt*RN4|@RvfAvWi*rCd}YsH!#`8&;;M~jy>Qr9I%3Ld2B#5e(j;Yy
zDb11XT|W5~ji9SO1l;_iBX>&b7vy(K2Hl25AGY7QgjpLLYw$NnU7twL`;S%l=)$?B
zTsg2)QK=nw?<0)|$Z}>nJ&$nGaKIPFJ;SR5Wr5{^fcXLT;@HsvR;e-+Xdq;?`m0i;2+vzhM+CN
z0<(oNoE}n}Hjw?KSHGzxD84*8Z}oZ4)XjIldwO)gp{(|opOhvou=IqAy#bj8SUB={
zV8*L{z6W4}IF7!f%XazhtQN}~PBT|UJrupOMbT`F6&AMIQ)w0enP>hTG{XLE$GH=l
z3dn$;kS4Q37`RL8B&(LPez?i7`?Fx|XNoY+y!EUizRp{d81J>}X99f3xF>R&>Jh`v
zw@xul)!*NM7Gr3jUevdfbnKMat
z8wq1`5Q0SgT}x8FFK~sfpK{YKS}pahRly$;?p?yvnf_co`6iUP|j_Cu}z;E79P&5ayTGF@3K>+CUtB^gQdHj`M7;wvo
zys(mb9h|L+wYx6q8=E-%n`zV9ZzCTVfQ9-yX^6Gg>pP}m8Un0~MWxa{%~4){Nc#3b
z>Feg@FoZ5OXhR>}%pVRoo}N_g1Mof{n-wG}$#*REubIpvP}%RCd>cg{CI=<$Rjs{<
zs|A$#)8O@^h=x{C&FH_UA~;YRBHirAHvRlbJK}Z`w!lw2OzgA27oroq=AFq
zc3Zi}(&wP(aS!$eBw$`$!VMv018(~%7xa}#$3cyXaKVk4<(@WB)7#i
zu_=z#8;}9bjs48%j*(@b6%_#ba4q-dg{F0oVgIWZz@(4I%uMLe_^VHPcH}i^0pS_o
zsgeOXj_q4!UrW7M**obg&f~T>T2Hdm7})v#-DWl+|I#QzxyURA;eOK&oL21ZA5fqAn0^wput4jvVR`Tjy*|5t!QlBGg~?bx
zFNulgrn58lHT!y3oBCR0U`|!`@M*5mu}AXrI~A&}oL|K55rP+44Q&yhW%MZ&fHkZw
zeXjJb*^W=a>vD&)D^9EL)rObwd-NK#-Yp-L9sT@i)?apT4DSS<#(-mX8_hkzO|41Y(Ho@g9l|iZAwes5$%zB
z)QgA>v0Pocnk_$5aI?k1QCIZ%Qk)5J1a^55#0W4hp2b{`lphu8n{7vwS-UveZD83t
zyYSZ6-$p?PbZw>oS{|Jay86e2Gflq5>WR;T02$u=58y-V33*hOhD?$lt)`4S15S4>
zNae__8!Aveq5%U#jlyJ&nyC(w?Pvh9z#;dc&=H
z)14AYo(B1Ti&3L?7x0o5?|h8tJXKsUFgsj>`94xXk71o{kGcEs=@Ju-$A^H2tqQx!
zAhgY_eKJ{e5S3pZ!&ntT2xu*}{yCXTI#KerO;B;O7&DAM&Ju=1wHgof#UwU_6rM_j
zu?L)_PHx1}KL#ORRO!RjJ&U89YMNo-7{qhgb*g2YnF7QG-2ANry-B?uNN_8_iL%
zqXOB)n-HU+&{~OpD}#qKbOOU!l0(Z5cS3E%RhXvXz%klW2xrEEr-IIK$VOu$3I-U!g?;UT3%iMg3#{B
zu_oqgJVKBI8<-C~o^}$l;9ChMqn7ebr6cB-6Z{-r0$*?{+
zd7rip?K$8MGHvx{pEZ{A5x@_)>dS@+e~QiB&{Ggv(EoLzHC)c%1&yTZtxWO(Ylj`w
z{g)xub}?hvz*)(WJgumt4-QW?5*1S{uu*B%mXa)LX&0|YykzMYY{jV&(sH+M5PZ)_dm7&+F+)a&5ygl^`!Fa8d`;
zhv#=#Klr;I&?uQCY*Cdm(Gq{Bl<=&^6_-}=VW=zEc@m|hwF#B3Jj!rLKA5!LDzBgD&VFBQ5p)g83NtO-c@Z-nZT9i%y(bsb9%=U}TRrnJ@@D8kF3BQdS{8|8pU#r0d+Id;YV#d`>g
zt-GPxl5h0(N
zCdD>yjFg~Os9G;JHQ5Gyh+(oOyAyb=R%dS2t;z11DQl=L10H|PXh8^wzK(U#tq7$`4Iq|1t%Q=k>fsGJj>?Hmx-Ot^5Z^@*WM5fBU0<==`l6eJgv_+X7ex
zZSge>qaPz6B77#DQPz4yaAR^yqGZile{p+BVWubgz}_V+=DaKMLroeq#+}vGBgUU*
zR<@OERW$)PK4v%HX9*f2*TY8YgXQntJKcq+PfjNUY<@eK?({nm3PIRW(w%O(Vx5{e
ze->wNJzgr=da8fW9N#_d<;@{fd9M-MNzE2DQ8;CB4yRSEMuQBBg26pJt%~=^wb>(u
z$*FsT{Tt`KP`yvX6Qzb98Fh88k(7O~4$
zTCcY*&ls$9n2Uy-EP=BK$;hc-iM95)HXwt-I$E2MQNx8zyE%-T>J|XWE0r$R=`7r4
zpIivZn5c~u)mRt2Es}VwHur|FeCr0-dog2){>Z>SBZEPvC9Zz5;`UsGOe*>0X@&9)
zle3(lk^0P}ki)rgNW8J=wbm9UN@z8C
zo`j*)qqJi>r^+A3gm|xgYE4!S8!0U1%{YqkKyM5^ao&$zbAFR#yTD5nwoe||%9A`v
zJsEF4Ae6cftlG}!ffE!x70+8GI;0GVXD^f;4^BCYtBL|!csewk>HTO;9pRhmaUWjH
zqP^L;!4J`u*KvF@9tD4uP^bN7@P!~m0~WX~g!2|%22C$W&$HkBlZrcs5Kog5t_24z
z2@xoOum$H;|8?<-@6NGz?aLY8FdF2;#7d+cQA4W2gg)*C5QX?#&W>{DbJd5GB;@}`
zMBRnzBDSh-qx?~G5mjekLhL+yhZvld=n8q!ummzN1((N`qO|7`AgbIez#cx?8PV_?
zCgwZTcL_SVC1}J@V_kYj#+yp_ip+tmuIAk*VS53~lfSyd`$1M78+VAAjx(jUB@sOSk2;yvIwq
zm~c;9SI&Fotx|Hkj!@;YB(7|5h@(y-WAReOO9wfkzXrY%S*sW+4e1%Pb1D#3Cwt;?
zF940Z(`i7G2=pTfv6Lq&I!vDF;1_z_h*)cV$Zr(lnjH{X8-)_3O|zw7*|4NqT-YH
zKrPhj{Y#jPZI1*x5s|;@zxP_{)7(St4|rOeta$v?AOhbZ7;E?l;CjiOimIhYpVJxh
zk$Rh|XLR-1z`IL*O4f$amQ8Rt+?JKyHh$4<&+P^rmOQX^dSu%AOXsbKT7>jJS(i@9oQ4-Ld3KgMzVzaLr}l1XG8>epcec8vm{|ga
zu3$`I{C_q?oCoZJOrF_I@t0E>E2h;1tPLtY=KNWgsi8an=@PxI=d+lAmz1TbYCb#H
zNxr%ne1_Q*rNpyz63%b~?#!vaS8pN!&7*L)5rh7KIV3LAnI!>RmG~m6_TP{WDs;cPVK+iXcXRL(3^7`3!Z2dfQZU*3@KE-VLI_K75O03HNVZc+}eitM7*;EhpF4cv_RHFkf
zL0uGsSHFi@?Q;q_
zU!Z#ar;!C{9A+3WwCzCK(D2_}bY3B;JZ1VYW&+(9^SYUlBI0!{P-kDxiIIuX|6>P%
zYaj{n4Njnop*y=E|8J}Cz3oU6$3ZP44HBY#s7t_&@wOlZtCSff6#!A^nKy}9|uuP4Q+S{E6wO(5#j>d6kG&SJ9mhDrRhupSGLY^N=SH;NcrjD6gYh1dPi
zkcsvlgrAO;PBYxs<6s|JjLO3G>2OcZ^u{^kp-9kmvVv@{4
z=GS{hKpjDK0-)6)PD+c8FDw)fT$r%v^5C8=#jDO7*59TWP|!PX5I|%T`Lk2KaL2K%
zMHC>BgC9++lGcmMecf~GiV*8(ifazIAf=+;SA$jxHySB5J`GvvBuEc)PXt}Db8b*h
zgO>;-N<*HQTer>byKJ%hmYFi(t$=Sa_yj%
z&>tahE6}I+KbuydLWm)-zp3%{VpEWvUbWI)VUJH;<`gI(1q{~WL~aF~Ce`n*@p!y;
zpaeGi0~+4Pie0GI=CZBiXheLZA+~E9))ZNJTwXib{blR8s%OnTvA|pl<$+_`_UN
zkbzmH%W+?WiY?r}J@W;865V`uy*|<)9{?cabLywnPJjTXSZ;ho!XdSV6t>Qwb6C~R
z#nQjm;|ydL9OtfN#Z(9NBmiX(Xjt8Ldp8?UIP)D@@fHjR)5)jo&mjRv`@H7~$>gN?
zr|@)~4vESW4ZI0H&ZWaer(yzyy8&EFi}&MDem)_h#+r(`KdtxIyRuxX=5KqO#P;VZ
zD%E`YV^}%?F?BY!Xhot6i*X~6%#
zj`9yaMrG_W1`6$8Yt~XIHEHVp54?aA`WsEdunNWiusCBq{-BI9u50_+cOXdPhyMeW
zXok+Wt6?WBng2&e?__e0%r(sYjzY)(yQhM>^Oor?2xmpgO~F634**sGMr7GA`C?+;
z#^np0X#aW%Vavf-KVGP@g6rFPOr-bs?
zU17vA6RkpqUl%Y3+k+A!b(pYeZ+^J9%O5mfomTXIH&_mEJE~lF*)S^}#a9nK5x<9A
z5*2N3oX62@8)g8N`}^9p!|vj$<^UzDGa;X!D0QmPCUEl&K6*M5wr^CeBqy`zm{v)#
z3^;ub@TTRm)y~2mqY7WpXG#A(6u2o4=r~iWRO?;T4HRp|xdWP8-Y?SeXss&Kmvgi3T#@9keQf_w~
zruS#ZxluiXeKixR-mU;P=MliBac)KKzL@{(T3@H-O#z5vQ%Ra{=UjgjkX?{m)cC#V
zl=8=9$hfD$TpWCeL2GVEHvXwA^tgeYI6Du?jHEx%sfGBcHEgHZ(kG|c@XZ4|L5NBZ
zYyb63?_0N24gz!!C9l
z-O%%F7hMIX?Eg0&pRwU)UDWqwB!(;Sg?$n3Ei0XnKOS-c^S)X|BUzv4?&@fnsKPLS
zAmo8aWW4qd(wDhz&C9mWn(W)z$-~FU57bRoE%vk`(ZSn~CO{NUEt;fY3;;D`dC#uP
zKj(S=2VVXE;1_!rH>YDH;iaey=3cZ0mLj-7KlU3xN+OmX!MKD?)R!LVkbt6dv$
zQenv$cLMbBLe-|$W4c!?$-jAFaua~YjSE0#rg|S~AXubLdGo8%R!yjL0fHSDIppJ8
zUH2_Mv%-uqj{dwAWWvSOXHkyF26>79{sqiG>mn9crYtOPx&VcKZ$7G$emkzsPr4PG
zRf$sG11bc*I1?t_a;M!uR4)MnSgf4knhdAOf<3a*lrioiAWeaAu+(itc!9XUuXX`L
zQqF!N_G8VSc{{z!{LNEU(hqb|Gz&8#w!AoMSFFr}`E2r#I&Wtel%t7%omgfuvNmL_
zjcOGDg-)80Y<0
zhQ&kR=Q^PN7nEWTl<@p$6@99qJH8mS!4kanTG2PDOw
zvGi{BWuS)ot%Ou>dV@(E-C0?qIS&rw*b0=HsVsEL0;}G_7Ok%K}VC
ziM#!!f&(5%dF*5X>d&;Wj<=L{8?Sg*Z)@->KJuw=?g{6VzeKs@DJ6802SeYnjSsS{
zeN;;ai%`5r|$wnuaEtJU0v
zq?{9nNzF%W_^Kk)&KQ4x;xQR&-JO|@>xacGT
zU4j$;1I$GrDQh>RxOUwjdnpzZnrOt+|0;@%mgCQ2ehbB?wF22jF9C}CC~(y~ab&!N
zs?h<6EXakp!>(Eu!^n4f@z1biBKy!0%C0@5HBCHp@OM5Q0Sit8PZbd8Gj1e??L}GU
zIuxXj4d*btuTN!BEjptpSK7PKna>_RD){vke_Vo*r!xsU+7h&Ha6@xZumF@>1fj}*
zL{$oaU|%j{xOWG?aI3A&z&bxHsxrJbB4rBnKueYGixDuu_;OhC_f#Y!Dk`?$9@zH_
z9lBTrah(H{EHC>BvCp~v@CL8~J(AxrV_(KHtSLZNmEZ}aD3PD$mLJSzI!=Ef9R)R^
z=MViN1^psE(&N6R7lUG^0uH;(MJS{ffl^ys0yp`W=UME+Zx;rB?<@~SFWCV|do=9-
z5yq$h&EcZ*;62a_)Juw_n`Z@z;H9fQ)-r`Xv~NqG?g8}f77kBI0r2I$B|Tez)&Z_J
zgEN10|1!o=&Lc!65l;~fs#9&GYBURm1zVYD9U34Lv;?{^Sdo^}Qfkp3TID4Sf|SpIqp%&{*~|pYX9GLYl(jA*mMEQuhshPx0%~SReA}
zL7{)m(UTzM@@MRC+yo$&L@lZ4;&%L6p7-rfRQwV&$d#k5u7VKx=?jy(J&(aBB;(1z
zQV2y+1l~90TOW^N0>QHUc0Wk3+v#OjGM=nLc}0U%+x}b6u%;@*S!YOv$roYhzj#r@
z2@_sRAsB|&wY0Vl*>$3}@{_n%RZ&V%vRK;T2*H8mvMb-qILylmFPB|oi8}=7!|!uN
zGbvNbLc7-^k@Dvj!$5UnFX(caqw6mFl;=;_?{{`@TKBZTbePpMB8)_)^F%gVc
z>3nuN6gvElOJ01?zp5lQc%lbZcD;9BNEobL?Y&VFAU}4hK_SSEMQ`OZ%ayD+p_)sW
zk6K@2xoPJk)E7qsRCd%?=}{*{LS3HS@H>SNSmFBCQS>)Q(WNj0M_X!UMn00aP#uTs
z(8$x)3*tXqR;>^)K3a?NJrhS0dkRBd0D9_rY~zrx3!-i|`<#plT`3)_<0-FIE`c#P
zsytYTQBxW}&GKCAY$xzyD9KNJGp?k5u)xH2pLd)enSUTL8er^}LwWe`I8Bv$jT!ef
z>a4G(YW5kody0si)`~luWzH=yCRKs}1m)cI))L%?gwD<0m9Z9me^JTT(dS=gkd#>O
zD9D>4W+#eHf8x?3t%C3K-c=GPX}~E&6hQg0h+c
ziw-wZPT>?Mp3ucN_X^1GY@nrlV6(1SnI>B8>zkubBq>lgASlm|IG_aP
zPX(z`0wE9qDJg<<0=L{h@SXY2d~d!p?}we;eQ$T(?(EKcpZPquX`Sl^NC^i+a1ZPj
zFL(Kndsn%I0Nqe4;#PQmUTF@9yRxAc?@xULTUwuqx8uo48|GhYcc59~%0)U~CoV4?
zF%LZg4|>0dZFf9*HFnEWj;2SyIIS{B@8feOnTgJu?^L+O6BS)cD^8;&wa%%_LC
z#v$bQhi*67@*?In0Y&sR4-2!93at$*wfJTrHp{oRR9*c#Jb3VDMnkhz48nf?cYI&$
z`?Dt$8#a>zB6{VqW>2>TwI*i@d&N~)U{_#eWnY
z&&tLvKar!XiwRV(I6KA=@o_x#+>UhJtyX3L?#8+9Fx>rm9dvhsXa*=t5RSWF;nh^a
zmXXP;$pIr3J!loqO*n~F@8leMFmk81ubX(mVGv{i6GTO;G=}xyQ5gTqOS23plq5bM?@_1HWv+f2BiJ
z{V}t(O$h5Ukwsg&#_`b1o_gT#%3S?M(AU#+;`QAI#Q8sNJ@zLjrSQ*`jxd<>Akoi+fVfZ*;A^-2|D{GAw1ZrlFHf&kmA0)b
zZ1ieLn!r#a$@e$-u`EI3qc10<)?OkIXQ`8V2*moHR|!GTIJxNBl^H$U3Z
z@w8N?OFm2w1c*paCv$rr%la()(3buLY_B%_cjm>5$5*F
z-?%fW=zW+%cH)Y74aB(B_C^>)!#|OYCJ%=xYr7D=oYgGb!^b9~xKFiXIH8ptRbj2gY1F7PklDeMs1bFn;-^H?ex2
zDeow@TO~uDGM5Ko1bN8>R~|zTizKJfs&$@G!-p^bqzkER$Cm;FLia#`a`E*1bB7US
zT~He)FQsPES9#d&tTp@?dg3qzI|E14?-0ZwtGgo!+RGA`{@z)-ZldgMAwliDj5b+DjZQ*m0
z#Hx%gLO;vGI?Apx{PjcUkkzLdp1*L
zOa$ve+SUqjV`{mdO?;POpZK!vtkU*ZZoV$NZ7{)HbmJVFKE@QTvq&uFYzKVwZ(*O3
zIXU~Xi!*A!FmuFlfZSF!DEW^abZTX+9=rgps
zirmxmqeVX+1E#JQ`d}7ChhA(He@;~HO`5I
zC4DM}mE3a!)y05X%@1w8>*8bD?PE~&TP((gN`HiYi(hKvW`l0&2wW^MSR%K$5!m7+
zoJfdz^0lekQ|1e~Yr)7YX;^HM8=UPOFGp+^Q#&Tm{3yulgvP~M35I$5o^(h_z{;zV
z4XT!;X!Ai5OU;5AGq07^DN8XKSC5AWk<(KOkMa%lbjrOIZnhGq7fw#fZH0LF@*CMn
z)6to%=zNo9(v}m97!;P)e#FXnx3t_T<*Ud+TN)DgP+?{1yS+k|dAI`aTX+$9Dpr}<5>@kuvVXaes40>F!q;R0ZyG2u>
zbtc(3lBqz_t(LzJ1%G-iKwST|_x|7fTmGB1`UhWNbB&EhZaOW5H!G@$bY#V?{(Qt3
zF&BBF5`+$69zy#ZKCfmcn;kl9X7dezWn}>Ymhdt;<^_ftfEOfQ5WEUz_!_PZp;`U;
zp^f{Xy#%0EFCChHZJ4`)XP>@(1rUI*^zFWVkVc0JjR4+n+#2%EU_vrm=XN*HMI!JW
zM5F+y1*KL*J$@j_L_j;0X$$dkgSo^EW@02A4EFV}&-5s&Wj7%XN&#Zz7TOsy%TGKL
zpYH50VCTP~{`AV77mr;J^d~id=Y5fJ1;MG6GrDT0>KjwWyOHokKqzD*H9Y`0|Kj-{^HMA{~
Z)oQMz_*wl1sEjD!XnXO;T5G?gzX7<3XC(jt

diff --git a/docs/articles/xspliner.html b/docs/articles/xspliner.html
index af7af94..a6897eb 100644
--- a/docs/articles/xspliner.html
+++ b/docs/articles/xspliner.html
@@ -88,7 +88,7 @@
       
    Basic theory and usage
    
                         
    Krystian Igras
    
             
-            
    2019-08-31
    
+            
    2019-09-05
    
       
       
       
@@ -150,29 +150,29 @@ 
    
 Defining formula
    
 
    This sections shows, how to build formula interpretable by xspliner package using Boston Housing Data from pdp package.
    
 
    Read the data
    
-
+
    data(boston)
+str(boston)
+#> 'data.frame':    506 obs. of  16 variables:
+#>  $ lon    : num  -71 -71 -70.9 -70.9 -70.9 ...
+#>  $ lat    : num  42.3 42.3 42.3 42.3 42.3 ...
+#>  $ cmedv  : num  24 21.6 34.7 33.4 36.2 28.7 22.9 22.1 16.5 18.9 ...
+#>  $ crim   : num  0.00632 0.02731 0.02729 0.03237 0.06905 ...
+#>  $ zn     : num  18 0 0 0 0 0 12.5 12.5 12.5 12.5 ...
+#>  $ indus  : num  2.31 7.07 7.07 2.18 2.18 2.18 7.87 7.87 7.87 7.87 ...
+#>  $ chas   : Factor w/ 2 levels "0","1": 1 1 1 1 1 1 1 1 1 1 ...
+#>  $ nox    : num  0.538 0.469 0.469 0.458 0.458 0.458 0.524 0.524 0.524 0.524 ...
+#>  $ rm     : num  6.58 6.42 7.18 7 7.15 ...
+#>  $ age    : num  65.2 78.9 61.1 45.8 54.2 58.7 66.6 96.1 100 85.9 ...
+#>  $ dis    : num  4.09 4.97 4.97 6.06 6.06 ...
+#>  $ rad    : int  1 2 2 3 3 3 5 5 5 5 ...
+#>  $ tax    : int  296 242 242 222 222 222 311 311 311 311 ...
+#>  $ ptratio: num  15.3 17.8 17.8 18.7 18.7 18.7 15.2 15.2 15.2 15.2 ...
+#>  $ b      : num  397 397 393 395 397 ...
+#>  $ lstat  : num  4.98 9.14 4.03 2.94 5.33 ...
    
 
    We’re going to build model based on formula
    
 
    cmedv ~ rm + lstat + nox
    
 
    So let’s build a random forest black box model, that we use as a base for further steps:
    
-
+
    boston_rf <- randomForest(cmedv ~ rm + lstat + nox, data = boston)
    
 
    Now let’s specify which variables should be transformed. In this example only nox variable will use random forest effect (PDP or ALEPlot).
    
 
    To indicate transformation use xs(nos) symbol.
    
 
    So we have:
    
@@ -201,17 +201,17 @@ 
    
 
    Below we will use PDP random forest effect, that returns 40 response points for nox variable. By checking ?pdp::partial we can see that grid.resolution parameter specifies predictor grid.
    
 
    Now we should just specify: xs(nox, effect = list(type = "pdp", grid.resolution = 40)).
    
 
    Let’s verify correctness of this parameters with the bare usage of pdp::partial:
    
-
+
    rf_effect <- pdp::partial(boston_rf, "nox", grid.resolution = 40)
+head(rf_effect)
+#>         nox     yhat
+#> 1 0.3850000 23.38131
+#> 2 0.3974615 23.40352
+#> 3 0.4099231 23.49511
+#> 4 0.4223846 23.62153
+#> 5 0.4348462 23.60978
+#> 6 0.4473077 23.69874
+nrow(rf_effect)
+#> [1] 40
    
 
    We got data.frame with predictor and response values containing 40 rows. So parameter should correctly specify our expectations.
    
 
    Remark
    
 
    Here we can see that response function is presented as data.frame with two columns:
    
@@ -253,40 +253,40 @@ 
    
 
    Having the formula defined, we almost have all the required data provided to build GLM. The only one left, that the approach requires is the black box model that is the basis of our resulting model.
    
 
    We will use here model_rf random forest model that was built before.
    
 
    The final step is to use xspliner::xspline to build the desired model.
    
-
    xp_model <- xspline(
-  cmedv ~ rm + lstat +
-    xs(nox, 
-       effect = list(type = "pdp", grid.resolution = 40), 
-       transition = list(k = 10, bs = "cr")),
-  model = boston_rf
-)
    
+
    xp_model <- xspline(
+  cmedv ~ rm + lstat +
+    xs(nox, 
+       effect = list(type = "pdp", grid.resolution = 40), 
+       transition = list(k = 10, bs = "cr")),
+  model = boston_rf
+)
    
 
    Lets check the model summary:
    
-
+
    summary(xp_model)
+#> 
+#> Call:
+#> stats::glm(formula = cmedv ~ rm + lstat + xs(nox), family = family, 
+#>     data = data)
+#> 
+#> Deviance Residuals: 
+#>      Min        1Q    Median        3Q       Max  
+#> -13.4873   -3.2935   -0.8747    1.9991   26.9926  
+#> 
+#> Coefficients:
+#>              Estimate Std. Error t value Pr(>|t|)    
+#> (Intercept) -34.29942    5.05680  -6.783 3.32e-11 ***
+#> rm            5.23681    0.41600  12.589  < 2e-16 ***
+#> lstat        -0.52504    0.04355 -12.057  < 2e-16 ***
+#> xs(nox)       1.31670    0.16255   8.100 4.20e-15 ***
+#> ---
+#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
+#> 
+#> (Dispersion parameter for gaussian family taken to be 26.81888)
+#> 
+#>     Null deviance: 42578  on 505  degrees of freedom
+#> Residual deviance: 13463  on 502  degrees of freedom
+#> AIC: 3106.2
+#> 
+#> Number of Fisher Scoring iterations: 2
    
 
    As the final model is just the GLM, the summary is also standard. One difference that we can see here is the formula call. It has missing xs parameters, and xs is treated as standard R function (previously it was just the symbol used in formula).
    
 
    More details about can be found in following sections:
    
 
      
@@ -304,7 +304,7 @@ 
      
 Use cases - see xspliner in action with examples
 
    
 
    Here we shortly show how the plot random forest effect and transition for nox variable:
    
-
    plot(xp_model, "nox")
    
+
    plot(xp_model, "nox")
    
 
    
 
 
diff --git a/docs/index.html b/docs/index.html
index 00ca215..3c6c9a4 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -87,9 +87,9 @@
     
    
 
    
 xspliner - Using surrogate black-boxes to train interpretable spline based additive models
    
-
    
+
    
 
    
-xspliner’s pipeline: model %>% xspline(…) and analyze
    
+xspliner’s pipeline: model %>% xspline(…) and analyze
    
 
    
 
 
    
diff --git a/docs/news/index.html b/docs/news/index.html
index 8f127c1..0f524dc 100644
--- a/docs/news/index.html
+++ b/docs/news/index.html
@@ -115,9 +115,22 @@ 
    Change log 
    
     
    
 
     
    
-    
    
+    
    
 
    
-xspliner 0.0.2 2018-12-21 
+xspliner 0.0.3 Unreleased 
+
    
+
      
+
    • Summary method extended with comparison statistics
      • 
+
    • Extended methods for auto-extracting model metadata (lhs, response)
      • 
+
    • Added comparison plot for factor responses
      • 
+
    • Documentation extended with new examples
      • 
+
    • Ability to specify glm options
      • 
+
    • Added more informative progress messages
      • 
+
    
+
    
+    
    
+
    
+xspliner 0.0.2 2018-12-21 
 
    
 
      
 
    • Specify parameters hierarchy
      • 
@@ -138,7 +151,8 @@ 
      
     
      
       
      Contents
      
       
     
      
   
    
diff --git a/docs/pkgdown.yml b/docs/pkgdown.yml
index f795d0b..b2fe7fe 100644
--- a/docs/pkgdown.yml
+++ b/docs/pkgdown.yml
@@ -1,4 +1,4 @@
-pandoc: 2.2.1
+pandoc: 2.7.3
 pkgdown: 1.3.0
 pkgdown_sha: ~
 articles:
diff --git a/docs/reference/build_xspliner.html b/docs/reference/build_xspliner.html
index bffdbbd..c0b9fe2 100644
--- a/docs/reference/build_xspliner.html
+++ b/docs/reference/build_xspliner.html
@@ -117,7 +117,7 @@ 
    Helper function for building GLM object with transformed variables.
    
 
     
    build_xspliner(formula, model, data, xf_opts = xf_opts_default,
   xs_opts = xs_opts_default, link = "identity", family = "gaussian",
-  env = parent.frame(), compare_stat = aic, control)
    
+  env = parent.frame(), compare_stat = aic, control, ...)
    
 
     
     Arguments
    
     
@@ -165,6 +165,10 @@ 
     Ar
       
    
+    
+      
+      
+    
    control
       
    Fitting settings. See glm.control.
    		
     
    ...
    Another parameters passed from chosen method. Not used.
    
     
    
 
 
diff --git a/docs/reference/index.html b/docs/reference/index.html
index e39a449..7381c47 100644
--- a/docs/reference/index.html
+++ b/docs/reference/index.html
@@ -112,7 +112,7 @@
     
    
       
    
         Reference
-        version 0.0.2.9003
+        version 0.0.3
       
    
     
    
 
diff --git a/docs/reference/summary.xspliner.html b/docs/reference/summary.xspliner.html
index 47b213e..39db124 100644
--- a/docs/reference/summary.xspliner.html
+++ b/docs/reference/summary.xspliner.html
@@ -222,78 +222,8 @@ 
    Examp
 #>   R^2:  0.9960708 
 #>   MSE Black Box:  0.02251746 
 #>   MSE Surrogate:  0.02989727 
    
-# Classification model
-data <- droplevels(iris[51:150, ]) # selecting only two species data
-iris.rf <- randomForest(Species ~ ., data = data)
-iris.xs <- xspline(iris.rf)
+# Classification model
 
-# Comparing summaries requires providing prediction function
-# Prediction as probability for success
-prob_rf <- function(object, newdata) predict(object, newdata = newdata, type = "prob")[, 2]
-prob_xs <- function(object, newdata) predict(object, newdata = newdata, type = "response")
-summary(iris.xs, model = iris.rf, newdata = data, prediction_funs = list(prob_xs, prob_rf))
    #> Models comparison 
-#>   1 - Max prediction normed-diff:  0.3908407 
-#>   R^2:  0.958412 
    #> Setting levels: control = versicolor, case = virginica
    #> Setting levels: control = versicolor, case = virginica
    #> Warning: An upcoming version of pROC will set the 'transpose' argument to FALSE by default. Set transpose = TRUE explicitly to keep the current behavior, or transpose = FALSE to adopt the new one and silence this warning. Type help(coords_transpose) for additional information.
    #> Warning: An upcoming version of pROC will set the 'transpose' argument to FALSE by default. Set transpose = TRUE explicitly to keep the current behavior, or transpose = FALSE to adopt the new one and silence this warning. Type help(coords_transpose) for additional information.
    #>   1 - Max ROC diff:  0.56 
-#>   1 - Mean ROC diff:  0.8443484 
    # Prediction as final category
-response_rf <- function(object, newdata) predict(object, newdata = newdata)
-response_xs <- function(object, newdata) {
-  y_levels <- levels(newdata[[environment(object)$response]])
-  factor(
-    y_levels[(predict.glm(object, newdata = newdata, type = "link") > 0) + 1],
-    levels = y_levels
-  )
-}
-response_rf(iris.rf, newdata = data)
    #>         51         52         53         54         55         56         57 
-#> versicolor versicolor versicolor versicolor versicolor versicolor versicolor 
-#>         58         59         60         61         62         63         64 
-#> versicolor versicolor versicolor versicolor versicolor versicolor versicolor 
-#>         65         66         67         68         69         70         71 
-#> versicolor versicolor versicolor versicolor versicolor versicolor versicolor 
-#>         72         73         74         75         76         77         78 
-#> versicolor versicolor versicolor versicolor versicolor versicolor versicolor 
-#>         79         80         81         82         83         84         85 
-#> versicolor versicolor versicolor versicolor versicolor versicolor versicolor 
-#>         86         87         88         89         90         91         92 
-#> versicolor versicolor versicolor versicolor versicolor versicolor versicolor 
-#>         93         94         95         96         97         98         99 
-#> versicolor versicolor versicolor versicolor versicolor versicolor versicolor 
-#>        100        101        102        103        104        105        106 
-#> versicolor  virginica  virginica  virginica  virginica  virginica  virginica 
-#>        107        108        109        110        111        112        113 
-#>  virginica  virginica  virginica  virginica  virginica  virginica  virginica 
-#>        114        115        116        117        118        119        120 
-#>  virginica  virginica  virginica  virginica  virginica  virginica  virginica 
-#>        121        122        123        124        125        126        127 
-#>  virginica  virginica  virginica  virginica  virginica  virginica  virginica 
-#>        128        129        130        131        132        133        134 
-#>  virginica  virginica  virginica  virginica  virginica  virginica  virginica 
-#>        135        136        137        138        139        140        141 
-#>  virginica  virginica  virginica  virginica  virginica  virginica  virginica 
-#>        142        143        144        145        146        147        148 
-#>  virginica  virginica  virginica  virginica  virginica  virginica  virginica 
-#>        149        150 
-#>  virginica  virginica 
-#> Levels: versicolor virginica
    response_xs(iris.xs, newdata = data)
    #>   [1] versicolor versicolor versicolor versicolor versicolor versicolor
-#>   [7] versicolor versicolor versicolor versicolor versicolor versicolor
-#>  [13] versicolor versicolor versicolor versicolor versicolor versicolor
-#>  [19] versicolor versicolor versicolor versicolor versicolor versicolor
-#>  [25] versicolor versicolor versicolor virginica  versicolor versicolor
-#>  [31] versicolor versicolor versicolor virginica  versicolor versicolor
-#>  [37] versicolor versicolor versicolor versicolor versicolor versicolor
-#>  [43] versicolor versicolor versicolor versicolor versicolor versicolor
-#>  [49] versicolor versicolor virginica  virginica  virginica  virginica 
-#>  [55] virginica  virginica  virginica  virginica  virginica  virginica 
-#>  [61] virginica  virginica  virginica  virginica  virginica  virginica 
-#>  [67] virginica  virginica  virginica  virginica  virginica  virginica 
-#>  [73] virginica  virginica  virginica  virginica  virginica  virginica 
-#>  [79] virginica  virginica  virginica  virginica  virginica  versicolor
-#>  [85] virginica  virginica  virginica  virginica  virginica  virginica 
-#>  [91] virginica  virginica  virginica  virginica  virginica  virginica 
-#>  [97] virginica  virginica  virginica  virginica 
-#> Levels: versicolor virginica
    summary(iris.xs, model = iris.rf, newdata = data, prediction_funs = list(response_xs, response_rf))
    #> Models comparison 
-#>   Mean predictions similarity:  0.97 
-#>   ACC Black Box:  1 
-#>   ACC Surrogate:  0.97
    plot(xs_iris, "Sepal.Length")
    @@ -238,23 +238,23 @@

    Examp #> #> Deviance Residuals: #> Min 1Q Median 3Q Max -#> -0.67881 -0.07692 -0.03106 0.09672 0.46649 +#> -0.67489 -0.07877 -0.03048 0.09804 0.46780 #> #> Coefficients: #> Estimate Std. Error t value Pr(>|t|) -#> (Intercept) -1.52021 0.29110 -5.222 6.02e-07 *** -#> xs(Sepal.Length) 0.33583 0.35190 0.954 0.3415 -#> xs(Petal.Length) 1.84254 0.38360 4.803 3.85e-06 *** -#> xf(Species)versicolor 0.07022 0.17859 0.393 0.6947 -#> xf(Species)virginica 0.39507 0.23856 1.656 0.0999 . +#> (Intercept) -1.40810 0.30407 -4.631 8.03e-06 *** +#> xs(Sepal.Length) 0.39342 0.36819 1.069 0.2871 +#> xs(Petal.Length) 1.67365 0.35387 4.730 5.28e-06 *** +#> xf(Species)versicolor 0.08582 0.17800 0.482 0.6304 +#> xf(Species)virginica 0.41758 0.23705 1.762 0.0802 . #> --- #> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 #> -#> (Dispersion parameter for gaussian family taken to be 0.03105335) +#> (Dispersion parameter for gaussian family taken to be 0.03102474) #> #> Null deviance: 86.5699 on 149 degrees of freedom -#> Residual deviance: 4.5027 on 145 degrees of freedom -#> AIC: -88.211 +#> Residual deviance: 4.4986 on 145 degrees of freedom +#> AIC: -88.349 #> #> Number of Fisher Scoring iterations: 2 #>

    plot(xs_iris, "Sepal.Length")
    @@ -269,7 +269,7 @@

    Examp #> -> data : 150 rows 4 cols (extracted from the model) #> -> target variable : not specified! (WARNING) #> -> predict function : yhat.randomForest will be used (default) -#> -> predicted values : numerical, min = 0.1991898 , mean = 1.199876 , max = 2.134114 +#> -> predicted values : numerical, min = 0.1977761 , mean = 1.199116 , max = 2.143874 #> -> residual function : difference between y and yhat (default) #> A new explainer has been created!

    xs_iris <- xspline(rf_iris) summary(xs_iris)
    #> @@ -279,23 +279,23 @@

    Examp #> #> Deviance Residuals: #> Min 1Q Median 3Q Max -#> -0.67881 -0.07692 -0.03106 0.09672 0.46649 +#> -0.67489 -0.07877 -0.03048 0.09804 0.46780 #> #> Coefficients: #> Estimate Std. Error t value Pr(>|t|) -#> (Intercept) -1.52021 0.29110 -5.222 6.02e-07 *** -#> xs(Sepal.Length) 0.33583 0.35190 0.954 0.3415 -#> xs(Petal.Length) 1.84254 0.38360 4.803 3.85e-06 *** -#> xf(Species)versicolor 0.07022 0.17859 0.393 0.6947 -#> xf(Species)virginica 0.39507 0.23856 1.656 0.0999 . +#> (Intercept) -1.40810 0.30407 -4.631 8.03e-06 *** +#> xs(Sepal.Length) 0.39342 0.36819 1.069 0.2871 +#> xs(Petal.Length) 1.67365 0.35387 4.730 5.28e-06 *** +#> xf(Species)versicolor 0.08582 0.17800 0.482 0.6304 +#> xf(Species)virginica 0.41758 0.23705 1.762 0.0802 . #> --- #> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 #> -#> (Dispersion parameter for gaussian family taken to be 0.03105335) +#> (Dispersion parameter for gaussian family taken to be 0.03102474) #> #> Null deviance: 86.5699 on 149 degrees of freedom -#> Residual deviance: 4.5027 on 145 degrees of freedom -#> AIC: -88.211 +#> Residual deviance: 4.4986 on 145 degrees of freedom +#> AIC: -88.349 #> #> Number of Fisher Scoring iterations: 2 #>

    plot(xs_iris, "Sepal.Length")
    diff --git a/man/summary.xspliner.Rd b/man/summary.xspliner.Rd index 5069391..e33a0a9 100644 --- a/man/summary.xspliner.Rd +++ b/man/summary.xspliner.Rd @@ -84,26 +84,5 @@ summary(iris.xs, "Species") summary(iris.xs, model = iris.rf, newdata = data) # Classification model -data <- droplevels(iris[51:150, ]) # selecting only two species data -iris.rf <- randomForest(Species ~ ., data = data) -iris.xs <- xspline(iris.rf) - -# Comparing summaries requires providing prediction function -# Prediction as probability for success -prob_rf <- function(object, newdata) predict(object, newdata = newdata, type = "prob")[, 2] -prob_xs <- function(object, newdata) predict(object, newdata = newdata, type = "response") -summary(iris.xs, model = iris.rf, newdata = data, prediction_funs = list(prob_xs, prob_rf)) -# Prediction as final category -response_rf <- function(object, newdata) predict(object, newdata = newdata) -response_xs <- function(object, newdata) { - y_levels <- levels(newdata[[environment(object)$response]]) - factor( - y_levels[(predict.glm(object, newdata = newdata, type = "link") > 0) + 1], - levels = y_levels - ) -} -response_rf(iris.rf, newdata = data) -response_xs(iris.xs, newdata = data) -summary(iris.xs, model = iris.rf, newdata = data, prediction_funs = list(response_xs, response_rf)) } diff --git a/vignettes/automation.Rmd b/vignettes/automation.Rmd index ae692c3..b4424d8 100644 --- a/vignettes/automation.Rmd +++ b/vignettes/automation.Rmd @@ -206,7 +206,7 @@ They are: - Building the black box with formula - Black box training data stores factors in long form (the factor is one column, not spread into binary wide form) -Most of R based ML models satisfy the above condition. One of exceptions is XGBoost which in actual xspliner version needs to be handled in the more complex xspline call. You can see it in [use cases](./use_cases.html) +Most of R based ML models satisfy the above condition. One of exceptions is XGBoost which in actual xspliner version needs to be handled in the more complex xspline call. You can see it in [use cases](./cases.html) ## Excluding predictors from transformation