[Chore] Cache extension coverage by ownership

[zoomote]

​Created by Roomote. Follow up by mentioning @roomote, in the web UI, or in Discord.

What changed

• Split extension unit coverage into independently cached API, core, services, misc/platform, and tree-sitter lanes.
• Give ownership lanes complete upstream workspace-build dependencies and conservative shared-production inputs while excluding other lanes' test files.
• Enforce exact resolved Vitest collection partitioning and the verified LCOV population in the coverage contract.
• Run at most two coverage processes concurrently after profiling showed five-way execution causes resource-contention failures.
• Preserve the same-commit/OS serialization that prevents duplicate Windows runs racing the Turbo cache save.

Why this change was made

The prior tree-sitter/general split left 414 general test files coupled to one cache key. Profiling found stable ownership boundaries of API, core, services, and misc tests, while production import cones overlap heavily. The implementation therefore isolates test-only invalidation but conservatively reruns all ownership lanes for shared production changes rather than risking false cache hits.

Impact

The five lane configs resolve to exactly the monolithic unit population: 478 discovered files, 8,411 tests, 469 LCOV source records, and 30,229 instrumented lines.

Measured isolated local cold times are API 16.2s, core 48.0s, services 36.6s, and misc 31.5s. The complete five-lane graph with tree-sitter completes in 1m56 cold and 310ms warm. An API-test-only change completes in 17.7s with five of six Turbo tasks restored; an API production change conservatively invalidates all four ownership lanes while retaining the tree-sitter cache.

Hosted Windows run 34793268975 completed successfully in 7m06s total. Its cold extension coverage step took 4m40s, down from 6m55s and 8m58s on the earlier two-lane cold runs (32% and 48% faster). The exhaustive coverage contract took 12s. A hosted partial-cache run was not manufactured with a no-op test edit; the deterministic local partial run is the measured evidence for that path.

Closes #115

[coderabbitai]

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Advanced

Run ID: 8a3b0ae9-e027-4eb9-9ca7-d43d39f3357a

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

Review status

This PR was opened by an automated account. A human maintainer must verify the change intent, provenance, and validation before merging.

Current step: Required CI passed. Waiting for automated review of the latest commit.

If automated review does not start, a maintainer must restart it.

Review-state labels are managed by this workflow; do not edit them manually.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!