Raise FileExistsError if colabfold raw directory is found before performing colabfold alignments

[jnwei]

Summary
The existance of the raw colabfold directory, in particular, the raw/out.tar.gz file from a previous run can cause the colabfold pipeline to assume that the colabfold job has been previously completed and skips submitting a new job.

In the worst case scenario, if a previous inference job crashes, and a new one is submitted, the wrong alignments from the outdated out.tar.gz file will be parsed and create potentially silent issues for prediction.

This PR raises an FileExistsError in the case that the colabfold raw directory still exists, and instructs the user to delete the directory. At this time, we leave the task of deleting the directory on the user in case they want to inspect or save the raw files.

Changes

• In preprocess_colabfold_msas add an error if the raw/ output directory exists
• Add a new test TestColabFoldQueryRunner.test_preprocess_raises_if_raw_dir_exists to check for FileExistsError
• Updates two previous tests as a result of the check:
  • TestColabFoldQueryRunner.test_empty_m8_file_handling : Replace the full preprocess_colabfold_msas call with add_msa_paths_to_iqs. I believe the intended purpose is to check the assignment of the templates alignments / entry ids upon receiving an empty template pdb70.m8 file. The rest of preprocess_colabfold_msas is already performed separately in this test in order to add additional assertions and checks.
  • TestColabFoldQueryRunner.test_features_on_multiple_queries_with_same_name: Delete the raw directory upon completion of the test loop.

Related Issues
#39

Testing

• Passes all tests in openfold3/tests/core/data/tools/test_colabfold_msa_server.py

Other Notes

• We may still want a more extensive fix as proposed in Preserve Raw ColabFold MSA Output Files #35. This is a more minimal fix to help navigate around the footgun that arises in Always delete raw Colabfold folder before processing #39

[jnwei]

TestColabFoldQueryRunner:test_empty_m8_file_handling: Replace the full preprocess_colabfold_msas call with add_msa_paths_to_iqs. I believe the intended purpose is to check the assignment of the templates alignments / entry ids upon receiving an empty template pdb70.m8 file. The rest of preprocess_colabfold_msas is already performed separately in this test in order to add additional assertions and checks.

@qurat-ul-ain95 Could you verify that the above change to the test test_empty_m8_file_handling is correct. I think I have understood the logic of that test, but I'd appreciate your validation as well.