feat: integrate AIGuard diagnosis evidence into Lab

[hyeokjun32]

Summary

• Integrated the AIGuard diagnosis report contract into Lab compare/report/deployment decision flows.
• Added guard_verdict normalization while preserving legacy status-based Guard payloads.
• Added Markdown/HTML rendering for diagnosis source and evidence items.
• Propagated Guard source paths into optional AIGuard compare input.

Validation

• PYTHONDONTWRITEBYTECODE=1 poetry run python3 -m pytest tests/test_deployment_decision.py tests/test_report_generators.py tests/test_compare_service.py tests/test_api_response_contract.py tests/test_worker_contract.py tests/test_api.py tests/test_compare_latest.py -q
• PYTHONDONTWRITEBYTECODE=1 poetry run python3 -m pytest -q

Impact

• Keeps Lab as the final deployment decision owner while allowing AIGuard to provide optional deterministic diagnosis evidence.
• Preserves existing legacy Guard status contracts and adds support for guard_verdict-based reports.

Notes

• No production SaaS, DB/queue/auth/billing behavior added.
• AIGuard remains optional; skipped or absent evidence remains supported.