Repair v0.9 semantic decoy coverage gates

[AbdelStark]

Parent #385
Closes #389

Summary

• add execution-decoy coverage summaries with generated, pack-level, candidate, scorable, and missing-record counts by category
• block semantic score gates with typed semantic_decoy_pair_count_blocker reasons when scorable pair minimums are not met
• tighten semantic decoy rows to preserve same-source benchmark/problem alignment and record schema metadata
• document pre-score coverage reporting in the API, observability spec, and benchmark template

Validation

• uv run pytest tests/eval/test_execution_surprise_decoys.py tests/eval/test_execution_surprise_claim_gates.py tests/eval/test_semantic_decoy_pack.py tests/eval/test_execution_eval_cli.py tests/docs/test_execution_substrate_docs.py tests/docs/test_release_templates.py
• uv run python -m compileall -q -x 'tests/fixtures/codestate/invalid_(before|after)\\.py$' codelewm tests scripts/build-passfail-pack\n- uv run pytest tests/\n- git diff --check\n