Fix CI/CD false positives and registry flakes

[konard]

Fixes #146.

Summary

• Add a tested 30 second default timeout to the shared JS test runner so Bun does not apply its 5000 ms default to integration-style tests.
• Harden the Rust workflow against transient crates.io HTTP/2 registry flakes with Cargo network retry/multiplexing settings.
• Upgrade Rust workflow cache steps to actions/cache@v5 and commit rust/Cargo.lock so the existing lockfile cache key is real.
• Preserve the requested CI logs, metadata, template snapshots, research notes, reproducer logs, and verification logs under docs/case-studies/issue-146/.

Reproduction

• Downloaded and analyzed JS CI run 28246415576; failures were Bun per-test timeouts at 5000 ms in Windows publish-to-crates tests and Ubuntu screen isolation output.
• Downloaded and analyzed Rust CI run 28246415639; build failed while Cargo downloaded wasm-bindgen-shared with [16] Error in the HTTP2 framing layer.
• Added js/test/run-js-tests.mjs; before the runner refactor, the focused test could not import the timeout policy, and after the fix it verifies default and explicit timeout behavior.

Verification

Logs for these checks are committed in docs/case-studies/issue-146/data/.

• cd js && bun install
• cd js && bun ../scripts/run-js-tests.mjs test/run-js-tests.mjs test/publish-to-crates.mjs
• cd js && bun ../scripts/run-js-tests.mjs test/cli.js
• cd js && bun run test
• cd js && bun run test --coverage --coverage-reporter=text (689 passed, 86.33% line coverage)
• cd js && bun run lint
• cd js && bun run format:check
• cd rust && cargo fmt --all -- --check
• cd rust && cargo clippy --all-targets --all-features
• cd rust && cargo test --all-features --verbose
• cd rust && cargo test --doc --verbose
• cd rust && cargo build --release --verbose
• cd rust && cargo package --allow-dirty
• bash scripts/check-mjs-syntax.sh
• node scripts/check-file-size.mjs
• PR-diff scripts/detect-code-changes.mjs and scripts/validate-changeset.mjs

Template Follow-Up

The JS template already has the 30 second Bun timeout. The Rust template already has cache v5 and a Cargo.lock guard, but shares the Cargo HTTP/2 registry-flake exposure, so I filed:

• Harden Cargo registry downloads against transient HTTP/2 framing failures rust-ai-driven-development-pipeline-template#83

[konard]

Working session summary

Implemented and pushed the fix on issue-146-5fd1b386b75b.

PR is ready: #147

What changed:

• Added a default --timeout 30000 to scripts/run-js-tests.mjs, while preserving explicit caller timeouts.
• Added regression tests for the JS test runner timeout handling.
• Hardened Rust CI against transient Cargo registry HTTP/2 failures with Cargo retry/multiplexing settings.
• Upgraded Rust workflow cache actions to actions/cache@v5.
• Added and tracked rust/Cargo.lock.
• Preserved the full investigation, logs, template comparison, and root-cause writeup under docs/case-studies/issue-146/.
• Filed upstream template follow-up: Harden Cargo registry downloads against transient HTTP/2 framing failures rust-ai-driven-development-pipeline-template#83

Verification:

• Local JS checks passed: targeted tests, full bun run test, coverage, lint, format check.
• Local Rust checks passed: fmt, clippy, tests, doctests, release build, package.
• Post-push GitHub Actions passed:
  • JavaScript CI/CD: https://github.com/link-foundation/start/actions/runs/28252324715
  • Rust CI/CD: https://github.com/link-foundation/start/actions/runs/28252324716

Working tree is clean, and PR #147 is no longer draft.

---

This summary was automatically extracted from the AI working session output.

[konard]

🤖 Solution Draft Log

This log file contains the complete execution trace of the AI solution draft process.

💰 Cost estimation:

• Model: GPT-5.5
• Provider: OpenAI
• Public pricing estimate: $14.761045

📊 Context and tokens usage:

• 471.5K / 200K (236%) input tokens, 41.9K / 128K (33%) output tokens

Total: (471.5K + 8.2M cached) input tokens, 41.9K output tokens, $14.761045 cost

🤖 Models used:

• Tool: OpenAI Codex
• Requested: gpt-5.5
• Thinking level: max (~31999 tokens)
• Model: GPT-5.5 (gpt-5.5)

📎 Log file uploaded as Gist (5401KB)

• View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.

[konard]

🎉 Auto-merged

This pull request has been automatically merged by hive-mind.

• All CI checks have passed

---

Auto-merged by hive-mind with --auto-merge flag