TODO

Add option to load/save data using new BinnedData virtual methods.

Add structured (json?) logging of analysis outputs.

Add python scripting of analysis.

Add option for unweighted bootstrap trials, using combined covariance for all samples.

Test/fix jackknife resampling.

Create PkCorrelationModel.

Implement likelihood refit marginalization of nuisance parameters.

Add peak broadening parameter.

Dump results of XiCorrelationFit in some format to allow plotting like data.

Implement output of correlated errors, with sharing of off-diagonal contributions.

Implement options for integrating theory over bins with finite size.

Re-implement applyTheoryOffsets for pseudo no-BAO mocks.

Cleanup of anisotropic BAO model:
 - add hexadecapole to multipole model
 - use GSL interpolator derivatives
 - implement priors on a,b
 - infer whether to use a,b or scale based on errors?

Add alpha-beta parameter to XiCorrelationModel.