build(deps): upgrade to vitest 5

[trvon]

Summary

Completes the vitest 5 upgrade that #119 held back (closes the migration deferred from #106 and #108). Moves vitest and @vitest/coverage-v8 to 5.x and adapts the benchmark tooling to the redesigned API.

Vitest 5 removed:

• the module-level bench() export (it now lives on the test context: test("...", ({ bench }) => ...)),
• the built-in benchmark result table,
• the --outputJson and --compare CLI flags.

Changes:

• benchmarks/core.bench.ts rewritten against the context bench() API. Timing (time/warmupTime) moved from per-benchmark options to the run options passed to .run().
• benchmarks/report.ts (new) holds pure, tested helpers for the results table, the baseline JSON, and the throughput comparison — the formatting vitest 5 no longer provides.
• bench:baseline / bench:compare now select mode through BENCH_BASELINE_OUT / BENCH_BASELINE_IN env vars instead of the removed flags. The baseline still lands at .artifacts/benchmarks/baseline.json and the operator workflow is unchanged.
• test/benchmark-report.test.ts (new) covers the report helpers, including the new-benchmark and zero-baseline edge cases in the comparison math.
• docs/TESTING.md documents the three bench scripts.

Also gives the never persists more fires than the selected controller cap property test a 60s timeout (same fix as #119) — it does real store I/O and intermittently exceeded the 15s default under parallel load. If #119 merges first, this hunk is identical and rebases cleanly.

Test plan

• npm run test:coverage — 67 files, 1066 tests pass; coverage above thresholds
• npm run typecheck, npm run lint (2 pre-existing warnings), npm run build
• npm run test:package, npm run test:property
• npm run bench, npm run bench:baseline, npm run bench:compare all produce output
• npm audit --audit-level=moderate — 0 vulnerabilities

Note

Benchmarking is marked experimental in vitest 5, so the report.ts helpers deliberately depend only on the stable tinybench statistics (latency/throughput means, rme), not on vitest's evolving benchmark types.

[Copilot]

Copilot review overview

🟢 Approval recommended

Reviewed changes have no unresolved blocking issues.

Review effort: Lite
Findings: None

What changed in this PR

Upgrades Vitest and coverage tooling to 5.x, adapting benchmark execution and reporting to the redesigned API.

Changes:

• Migrates benchmarks to Vitest 5’s context-based API.
• Adds tested reporting, baseline, and comparison helpers.
• Updates dependencies, documentation, scripts, and a property-test timeout.

File	Description
test/​property/​store.property.test.ts	Increases the I/O-heavy property-test timeout.
test/​benchmark-report.test.ts	Tests benchmark reporting and comparison behavior.
package.json	Upgrades Vitest dependencies and benchmark scripts.
package-lock.json	Locks updated dependency versions.
docs/​TESTING.md	Documents benchmark commands and output.
benchmarks/​report.ts	Provides benchmark reporting and comparison helpers.
benchmarks/​core.bench.ts	Uses the Vitest 5 benchmark context API.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[trvon]

Restacked: this PR now targets build/dependabot-batch-2 (#119) instead of master, so its diff is the vitest 5 change alone. Merge #119 first, then this. The shared store-property timeout fix now comes from #119 and is no longer duplicated here.