Adding tests to cover the CFEL method's outputs

[hverdonk]

Summary

• Implemented more comprehensive CFEL method tests that cover improved per-site data extraction and formatting features introduced in conserved_aa now counted correctly, conserved_nt removed #9 and Calculate site aa composition & substitutions on a per-comparison-group basis #11.
• Fixed beta index mapping
• Improve determinism of existing test results by enforcing sorted combined outputs.
• Bump version to 0.1.3.

New tests

• Add tests/test_cfel_Beta_and_Qvalue_updates.py:
  • Validates diff_sites from Q-value threshold (≤ 0.20)
  • Validates per-site/group cfel_marker and cfel_beta formatting
  • Confirms expected comparison-site field names from CfelMethod.get_comparison_group_site_fields()
• Add tests/test_cfel_comparison_outputs.py:
  • Verifies group_N, group_T, group_dN/dS, group_aa_conserved are present in the output csv combined_comparison_summary
  • Asserts presence/types of composition, substitutions, majority_residue
• Replace old integration test data with new integration test data that includes a substitutions field
  • All other fields in the new integration test data are identical to the fields in the old test data, so the new test data is backwards compatible.
• Update tests/test_cli.py to ensure deterministic checks (sorting output CSVs to ensure that the first gene in the output matches the first output item checked in the assertions).

Version

• drhip/version.py bumped to 0.1.3.

[d-callan]

i like this. only thing id like to see is that were still backward compatible to older cfel results json without the subs info. i think all that means for this pr is that rather than replace old tests json files, we add new ones and then write tests conditionally. possibly drhip should produce a warning for older json.

[hverdonk]

I like the idea of DRHIP producing a warning for an older json without the 'substitutions' field. All the other json fields are identical between old and new jsons, so backwards compatibility shouldn't be an issue. I'm happy to add old versions of the json test files + conditional tests if you'd prefer, though.

[d-callan]

yea the concern is to make sure we know rather than assume we wont err if were missing certain 'expected' fields. too easy for someone (me, 6 months from now when i dont remember this pr) to modify something that may break this assumption. other fields we may want to err if missing..

[hverdonk]

Okay, I've added some results json field validation code. If you like it, I'm happy to merge what we've got into the main branch

[d-callan]

Being explicit about required fields makes me happier, thanks. Looks like we included cfel subs in the required list though? And this doesn't really protect us against the case where someone introduces some logic that will err if the subs field is missing even if it's not explicitly declared required. But merge if you like, and I can always look at it later too.

[hverdonk]

A lot of the internal code already checks to see if required fields are present before trying to calculate, for example, majority residue. If a required field isn't present for calculating a given thing, DRHIP warns the user that the required field isn't present in the results json, and that the thing isn't present in the output. For me, this covers the case where we're using old CFEL results that don't have the substitutions field - DRHIP will still produce what output it can, but it will warn that one of the required fields is missing from CFEL and skip a few fields as a result.

If we wanted to be stricter, we could always force DRHIP to quit with an error if a required field isn't present in the input data, or add a --strict command line option where warnings lead to an early termination instead of letting DRHIP analysis continue