Add MLflow logs to evaluate job

[sfmig]

What does this PR do?

• Links evaluation job to corresponding training job
  • Adds info of the trained model being evaluated in MLflow
  • Removes option to set the evaluation job run name as a function of the training job run name (mlflow_run_name_auto is now the default behaviour)
• Sets evaluation to run by default on validation split (rather than test split)
  • Adds an option to run it on the test split with a CLI flag
• Changes name of directory with evaluation output
• Updates README

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 11.11111% with 32 lines in your changes missing coverage. Please review.

Project coverage is 46.77%. Comparing base (7105c4c) to head (a0f7f84).

Files with missing lines	Patch %	Lines
crabs/detector/evaluate_model.py	0.00%	25 Missing ⚠️
crabs/detector/utils/evaluate.py	12.50%	7 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #220      +/-   ##
==========================================
- Coverage   47.69%   46.77%   -0.92%     
==========================================
  Files          24       24              
  Lines        1495     1520      +25     
==========================================
- Hits          713      711       -2     
- Misses        782      809      +27     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.