FIX TrueFalseResponseHandler: strip whitespace before validating verdict

[WatchTree-19]

Description

TrueFalseResponseHandler.parse lowercases the parsed score value but does not strip it before checking membership in {"true", "false"}:

normalized_value = score.raw_score_value.lower()
if normalized_value not in {"true", "false"}:
    raise InvalidJsonException(...)

raw_score_value comes straight from str(parsed_response[score_value_output_key]), so a judge that returns a valid verdict with incidental surrounding whitespace - "true\n", " false", "True " - is rejected as out-of-domain. Targets that do not natively enforce the JSON schema can easily emit this, and the result is that a usable true/false judgment is thrown away (and the JSON retry path is triggered, burning a call).

The numeric path already tolerates this, since float("3.0 ") succeeds; only the string-compared true/false domain is whitespace-sensitive.

Fix: strip() before lower() so incidental whitespace no longer invalidates an otherwise-valid verdict. The stored raw_score_value remains the clean "true"/"false".

Same class as #2133 (parse the raw score robustly before validating it).

Tests and Documentation

Added a parametrized regression test in tests/unit/score/test_response_handler.py covering "true ", " false", "True\n", and " FALSE " -> all now normalize to the expected verdict. Out-of-domain values (e.g. "refusal") are still rejected.

pytest tests/unit/score/test_response_handler.py tests/unit/score/test_self_ask_true_false.py tests/unit/score/test_general_true_false_scorer.py -> 46 passed. ruff and black clean. No documentation changes needed (internal parsing behavior only).

[WatchTree-19]

hannahwestra25 if you have a moment - this is the same small scorer-hardening class as #2388 (thanks again for merging that one). here the true/false response handler lowercases the verdict but doesn't strip it, so a judge returning a valid true/false with trailing whitespace (e.g. true\n from a target that isn't schema-enforced) gets rejected as out-of-domain and the score is dropped. fix is .strip().lower(), with a regression test. CLA's green - i think it just needs a maintainer to approve the CI run on the fork. no rush, and happy to point it at whoever's best.