Add option to load `Case` results with older verion of the project repo

[schmoelder]

When loading results using:

case = Case(study, options=options, name=name)
case.load()

the following error may occur:

Found matching options hash 1234abc, but incorrect study commit hash
needs: d96bx63, has: abcd12e

This error indicates that the study code has changed since the results were generated. However, in many cases, the changes are unrelated to the loaded results (e.g., modifications to other parts of the codebase). Currently, the process fails even if the results are still valid for the given options hash.

A workaround exists:

project_repo.copy_data_to_cache(results_branch)

But this requires knowing the results_branch, which is often inconvenient or impractical.

Add an option (e.g., allow_semi_correct_hits=True) to case.load() to return results even if the study commit hash does not match, as long as the options hash matches. This would allow users to proceed with potentially valid results, while still warning them about the mismatch.

[schmoelder]

Proposed solution:

• To OutputRepo class, add two new log dictionaries:
  • dict[branch_name, dict[project_commit_hash, list[options_hash]]]
  • dict[options_hash, dict[branch_name, list[project_commit_hash]]
    Note, this works for branches, but only if branch does not contain merges. Could be addressed later.
• In Case, overhaul _get_results_branch:
  • Add flag to accept older project repo versions
  • Add optional branch argument
• Adapt run_study() method (which internally calls has_results_for_this_run, which calls _get_results_branch).
• Also: add branch column to log.tsv and remove from ouptut_repo_commit_message. Also add datetime column (but keep it in branch_name) (see Add project_repo_branch_name to log.tsv #99)

[hannahlanzrath]

resolved by #101