test(performance): allow RVF search setup to complete

[rudycelekli]

Summary

The native RVF Performance Gates job can fail while preparing the search-latency benchmark: it ingests 1,000 patterns before measuring search, and the setup exceeded Vitest's default 10-second per-test timeout on #712's CI run. That run measured search p95 at 0.40 ms, below the existing 50 ms search bound, but reported a timeout failure. The separate ingest benchmark passed at 7.75 seconds and retains its strict 10-second throughput bound.

Give only the combined ingest-and-search test a 30-second execution budget so it can complete setup and evaluate the existing search-latency assertion. Neither the ingest threshold nor the search threshold changes.

Verification

• The linked GitHub job reproduces the timeout after roughly 15 seconds in the combined test; its measured search p95 was 0.40 ms.
• The real native benchmark passes locally: 4/4 tests, with 1,000-pattern ingest at 7.96 seconds and search p95 at 0.27 ms (Vitest 4.1.4). npm run typecheck and git diff --check pass.

Failure modes

• The combined benchmark's setup exceeds the framework timeout before its measured search assertion is evaluated. The linked CI job exercises this; the real benchmark passes with the explicit budget.
• A genuine ingest or search regression must still fail. The separate ingest test keeps ingestMs < 10000, and the combined test keeps p95 < 50 in tests/performance/rvf-pattern-store.test.ts.

---

Required check (issue #401)

• Every failure mode mentioned in this PR description has either (a) a test that exercises it, or (b) a linked tracking issue. "Unlikely" is not an acceptable substitute. If you wrote "I don't think this can happen but...", that sentence is a failure mode and needs a test or an issue link.

Optional context

• Linked issues: none; CI failure on fix(test-execution): reject incomplete Vitest execution evidence #712
• Trust tier change (if any): none
• Affects published API or CLI surface: no
• Touches the init flow / npm-publish.yml / tests/fixtures/init-corpus/: no

[proffesor-for-testing]

Thanks @rudycelekli — nicely scoped, and the linked #712 log backs it up exactly: search p95 was 0.40 ms and the only failure was setup running past Vitest's 10 s default. Raising the budget on just the combined test while keeping the strict < 10000 ms standalone ingest bound and p95 < 50 intact is the right trade. Verified locally (4/4 green) and CI is green.

Optional follow-up: logging the combined test's ingest duration would make future slow-runner diagnosis easier. Merging — thanks!