feat(bench): add RAG benchmark adapters

[drewstone]

Summary

• add agent-bench adapters for RAGBench, CRAG, NoMIRACL, Open RAG Bench, and T2-RAGBench
• add shared RAG answer/relevance parsing and deterministic scoring helpers
• add fixture-backed tests and public runBenchmarks coverage for the new adapters

Verification

• pnpm exec tsx --test src/benchmarks/rag-benchmarks.test.mts: 6/6 pass
• pnpm exec tsc --noEmit: pass
• pnpm exec tsx --test $(find src -name '*.test.mts' -print | sort): 50/50 pass
• fixture runBenchmarks smoke: 5 rows, 0 unavailable, 5/5 resolved
• git diff --check: pass
• adapter registry count: 31 total, 5 RAG keys present
• mergeability: git merge-tree --write-tree origin/main HEAD exit 0

[tangletools]

✅ Auto-approved drewstone PR — 47442094

This PR was opened by the trusted drewstone account.
The full PR reviewer audit still runs separately and will publish findings if it detects issues.

_{tangletools · auto-approval · reason: drewstone_author · 2026-07-07T22:33:50Z}