Slide improvements pass

[brookslogan]

Checklist

Please:

• Make sure this PR is against "dev", not "main" (unless this is a release
  PR).
• Request a review from one of the current main reviewers:
  brookslogan, nmdefries.
• Makes sure to bump the version number in DESCRIPTION. Always increment
  the patch version number (the third number), unless you are making a
  release PR from dev to main, in which case increment the minor version
  number (the second number).
• Describe changes made in NEWS.md, making sure breaking changes
  (backwards-incompatible changes to the documented interface) are noted.
  Collect the changes under the next release number (e.g. if you are on
  1.7.2, then write your changes under the 1.8 heading).
• See DEVELOPMENT.md for more information on the development
  process.

Change explanations for reviewer

Magic GitHub syntax to mark associated Issue(s) as resolved when this is merged into the default branch

• Resolves Consider allowing list-type output from epi[x]_slide inner computations #240
• Resolves [enh] promote other_keys  #446
• Resolves "Promote" other_keys to be printed, a constructor parameter, and more clearly documented #186
• Resolves epi_slide and epix_slide args #510
• Resolves (partially?) Consider dot-prefixing all parameter names in functions where we take in user-controlled named ... #162
• Addresses some (but not all) naming issues here Clarify naming and fix defaults: epix_slide is over versions #146

[brookslogan]

(Random implementation notes: I also investigated an as_slide_computations() approach that would give a length>=1 list of functions to apply in the tidyeval case and length-1 lists in the function and formula cases. Pros of that way include reducing code duplication/complexity/divergence when validating computation outputs [though perhaps we do want some differences, e.g., regarding interpretation of NULL (removing column vs. a missing element in a unnamed list of lists?)], [maybe reducing some naming and data-frame-combining-by-name overhead,] and having more freedom if we want to try to detect and optimize some slide computations by transforming them to calls to (internals of) epi_slide_opt() (as dplyr grouped_df's already do or did at some point). Cons of that way are that we may introduce some computational overhead in the function and formula cases (setting up, storing, munging results_envs --- this is assuming the loop over computations has been hoisted above the group and/or ref time values loops to enable other optimizations; if that's not the other case, then there's the overhead of repeatedly executing a loop over 1 function which might actually be relevant if you're doing an epi_slide() with an unrecognized simple function) + need for yet another argument (.data / results, not just .x --- either added user-side or via transformations which also have time/debuggability overhead) or code complexity to avoid this.)

[brookslogan]

@dshemetov I've updated epi_slide() with some of the changes I had planned, though it's probably harder to mess around with than an epix_slide() due to size rules / broadcasting. Maybe you could play around with the new behavior and see if it's desirable + take a look and see if named-list handling could fit in nicely? For the latter, you might need to modify: 1. epi_slide() around the slide_values_list validation, and potentially 2. as_slide_computation() in the tidyeval handling.

[dshemetov]

Perf has 3-4x slowdown due to validate_tibble and as_environment in as_slide_computation. To address later.

---

Perf comparisons of dev against this branch (branch a and branch b, respectively): profvis_slide_changes.zip, run with

# Profile
# Branch A
# epi_slide
p <- profvis::profvis({
  jhu_csse_county_level_subset %>%
    group_by(geo_value) %>%
    epi_slide(cases_sum5 = sum(.x$cases), before = 5)
})
now <- format(Sys.time(), "%Y-%m-%d %H_%M_%S %Z")
htmlwidgets::saveWidget(p, glue::glue("profvis_{now}.html"), selfcontained = TRUE)

# Branch B
# epi_slide
p <- profvis::profvis({
  jhu_csse_county_level_subset %>% epi_slide(cases_sum5 = sum(.x$cases), .window_size = 5)
})
now <- format(Sys.time(), "%Y-%m-%d %H_%M_%S %Z")
htmlwidgets::saveWidget(p, glue::glue("profvis_{now}.html"), selfcontained = TRUE)

[dshemetov]

Good to go. Will merge to main after this. epipredict PR will follow soon after.