Add helper to compute log_likelihood and stop computing it by default

[ricardoV94]

Closes #6266

Not sure about the function placement...

Major / Breaking Changes

• sample no longer returns the log_likelihood group by default. Explicitly set idata_kwargs={"log_likelihood": True} or call the helper compute_log_likelihood(idata) after sampling. This is needed when using arviz.compare

[codecov]

Codecov Report

Merging #6374 (888fe7b) into main (5692fc0) will increase coverage by 0.04%.
The diff coverage is 100.00%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #6374      +/-   ##
==========================================
+ Coverage   94.71%   94.76%   +0.04%     
==========================================
  Files         132      133       +1     
  Lines       26771    26825      +54     
==========================================
+ Hits        25357    25421      +64     
+ Misses       1414     1404      -10     

Impacted Files	Coverage Δ
pymc/tests/sampling/test_forward.py	100.00% <ø> (ø)
pymc/backends/arviz.py	96.37% <100.00%> (+5.65%)	⬆️
pymc/sampling/jax.py	98.19% <100.00%> (ø)
pymc/stats/log_likelihood.py	100.00% <100.00%> (ø)
pymc/tests/backends/test_arviz.py	99.02% <100.00%> (ø)
pymc/tests/sampling/test_jax.py	100.00% <100.00%> (ø)
pymc/tests/stats/test_log_likelihood.py	100.00% <100.00%> (ø)
pymc/backends/base.py	85.38% <0.00%> (-1.93%)	⬇️
pymc/stats/__init__.py

[OriolAbril]

I would not call the file model comparison. It is not doing any model comparison, it is only an intermediate result that can be used for that, and it is also not specific to model comparison, it is also used in loo_pit which is model checking similar to plot_ppc.

To add to docs, list is somewhere in https://github.com/pymc-devs/pymc/tree/main/docs/source/api. It doesn't really matter if it is "the wrong place" as #5282 is still open and the API docs need significant work.

Most important thing is the comment on dims and coords. If they are indeed preserved, we should add a test for that, if they are not we should fix it and add a test for that.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[OriolAbril]

Looks good, thanks!