Fix #90: render identity-threat framing in persona/reasoning context

[DeveshParagiri]

Summary

• Extend ReasoningContext with identity_threat_summary
• Add deterministic identity-threat detection in engine context building using scenario text plus agent identity attributes (political, religious, race/ethnicity, gender/sexual identity, parental role, citizenship)
• Inject a dedicated Identity Relevance prompt section so agents can explicitly reason when an issue feels identity-relevant
• Add tests covering context construction and prompt inclusion

Testing

pytest -q tests/test_engine.py::TestTokenAccumulation::test_build_reasoning_context_adds_identity_threat_summary tests/test_reasoning_prompts.py::TestPhaseAPromptFeatures::test_identity_relevance_included

Closes #90

---

Recreated from closed PR #97 (base branch was deleted)

---

⚠️ CHANGES REQUESTED - DO NOT MERGE

This PR uses hardcoded scenario keywords for identity detection:

if political_value and scenario_mentions(
    ("liberal", "conservative", "book ban", "school board", ...)
):

Problems:

1. Not configurable - can't add/remove keywords without code changes
2. Scenario-specific leakage - book ban, school board are from test scenario
3. False positives - men/man matches management, manual, etc.
4. Not extensible - new identity dimensions require code changes

Suggested fix: Add identity_dimensions field to scenario spec:

identity_dimensions:
  - dimension: political_orientation
    reason: "The policy is framed along partisan lines"

See full review: #97 (review)

[DeveshParagiri]

Superseded by #106 which implements a proper LLM-driven approach for identity_dimensions detection.