feat(bench): persist runtime decision points

[drewstone]

Summary

• record runtime semantic decision points in the benchmark hook recorder
• persist runtimeDecisionPoints beside runtimeEvents in bench corpus rows
• thread decision points through SWE/commit0/runLoop benchmark writers
• add corpus writer coverage proving decision points survive artifact creation

Why

Runtime hooks were emitting semantic decision points, but the benchmark recorder only captured lifecycle events. This keeps belief-state evidence in the same append-only corpus row as the run, without logs or side channels.

Verification

• pnpm --filter @tangle-network/agent-runtime-bench exec tsx src/corpus.test.mts
• pnpm typecheck
• pnpm lint (passes; existing warnings in tests/profiles/coder.test.ts)
• pnpm test

Note: pnpm --filter @tangle-network/agent-runtime-bench exec tsc --noEmit currently resolves TypeScript 4.5.4 and fails before this change on existing TS config/node type syntax; root pnpm typecheck passes.

[tangletools]

⚠️ Review Interrupted — 8fdf12b0

The review runner stopped before publishing a final verdict: webhook_restarted.

State	Detail
Interrupted	webhook restarted

No review verdict was produced for this run. Trigger a fresh review on the current PR head if the PR is still open.

_{tangletools · #191 · model: kimi-for-coding · updated 2026-06-07T21:16:35Z}