Revert "Revert "add default try catch to evaluator function""

[mayinghan]

Reverts #301

---

Note

Catches exceptions in evaluator functions to produce invalid zero-score results, preserves/updates statuses appropriately, deep-copies eval metadata, and adds comprehensive tests.

• Pytest evaluator:
  • Add try/except around execute_pytest in both pointwise and groupwise modes to set evaluation_result (score=0.0, is_score_valid=False, detailed reason) and mark eval_metadata.status = Status.error(...).
  • Only set eval_finished() if status is unset or still RUNNING; avoid overriding error states.
  • Deep-copy eval_metadata per row via eval_metadata.model_copy(deep=True).
• Postprocessing:
  • When is_score_valid is False, set Status.score_invalid() only if status is missing or not already an error.
• Exception config:
  • Minor cleanup of DEFAULT_RETRYABLE_EXCEPTIONS entries/trailing commas.
• Tests:
  • Add tests/pytest/test_pytest_evaluator_error_handling.py covering pointwise/groupwise exception handling, status codes, edge cases, and input_rows flow.

^{Written by Cursor Bugbot for commit 138d537. This will update automatically on new commits. Configure here.}

[dphuang2]

LGTM, thanks!