Python: fail assert_score_at_least when evaluator filter matches no scores - #10180

Fixes #10166.

Motivation & Context

Previously, assert_score_at_least() could silently pass when an evaluator filter didn't match any scores. This could happen because of a typo in the evaluator name or a missing evaluation check, making it appear that the score validation had succeeded even though nothing was actually checked.

Changes Made

• Updated EvalResults.assert_score_at_least() to keep track of matched scores and evaluator names across evaluation items and sub-results.
• Added a check to raise EvalNotPassedError when a specified evaluator doesn't match any scores.
• The error message identifies the requested evaluator and lists the available evaluator names to make debugging easier.
• Preserved the existing behavior when no evaluator filter is provided.
• Added regression tests covering unmatched evaluators, empty scores, custom error messages, and nested sub-results.

Validation

• uv run pytest packages/core/tests/core/test_local_eval.py — 97 passed.
• uv run pytest packages/core/tests/core/test_evaluation.py — 91 passed.
• uv run poe syntax -P core — Passed.
• uv run poe pyright -P core — 0 errors.

AI Assistance

AI assistance was used during issue investigation, implementation, and writing regression tests. All changes were reviewed and tested locally.