[P2] Guard verdict policy confidence coercion against non-finite values

[stranske]

Why

In scripts/langchain/verdict_policy.py lines 73-88, _coerce_confidence and _normalize_confidence do not validate math.isfinite(value). When a provider verdict reports a non-finite confidence string (such as "nan" or "inf"), float('nan') propagates into ProviderVerdict.confidence. In _split_pass_concerns lines 177-181, comparing max_confidence >= CONCERNS_NEEDS_HUMAN_THRESHOLD evaluates to False when confidence is NaN, silently bypassing the needs_human = True human-review trigger on split high-stakes provider verdicts.

Scope

• Update _coerce_confidence and _normalize_confidence in scripts/langchain/verdict_policy.py to validate math.isfinite and clamp invalid/non-finite confidence to 0.0.
• Add unit tests in tests/test_main.py verifying split verdicts with non-finite confidence values are safely handled.

Non-Goals

• Changing CONCERNS_NEEDS_HUMAN_THRESHOLD constant value (0.85).
• Modifying verdict classification mappings in VERDICT_SEVERITY.
• Scaffold-only completion does NOT count: modifying evaluate_verdict_policy without sanitizing _coerce_confidence and _normalize_confidence is a failure of this issue.

Tasks

• Add math.isfinite validation to _coerce_confidence in scripts/langchain/verdict_policy.py
• Add math.isfinite validation to _normalize_confidence in scripts/langchain/verdict_policy.py
• Add unit tests in tests/test_main.py covering non-finite confidence strings ("nan", "inf", "-inf")

Acceptance Criteria

• pytest tests/test_main.py passes with all tests green.
• python -c "from scripts.langchain.verdict_policy import _coerce_confidence, _normalize_confidence; assert _coerce_confidence('nan') == 0.0 and _normalize_confidence(float('nan')) == 0.0" succeeds.
• Deliberate-break demonstration: removing math.isfinite validation causes pytest tests/test_main.py to fail on non-finite confidence test cases, and reverting restores pass.

Implementation Notes

References: Code/Audits/Ready/2026-09-07-00-repo-map.md.
Ensure _coerce_confidence returns 0.0 for any unparseable or non-finite float value.