test: add issue 349 restart retest harness

[wbugitlab1]

Summary

• Add a dedicated GitHub Actions restart retest job for issue Lost all data after restart #349 on Ubuntu and macOS.
• Add scripts/github/issue-349-restart-retest.mjs, which starts the built CLI in a temp HOME and data directory, writes a sentinel memory through REST, flushes/stops through the supported CLI path, restarts, and verifies the sentinel survives.
• Keep production runtime code unchanged. This PR is evidence gathering for Lost all data after restart #349 after the All data lost on restart — stop order kills iii engine before worker can flush #338 / PR fix: checkpoint worker before native stop #1033 shutdown-flush fix.

Validation

• node --check scripts/github/issue-349-restart-retest.mjs
• corepack pnpm exec vitest run test/quality-gates.test.ts test/consistency.test.ts
• corepack pnpm exec vitest run test/index-persistence.test.ts test/search.test.ts test/shutdown-flush.test.ts test/api-boundary-coverage.test.ts test/cli-stop-port-detection.test.ts test/reconnect-registration.test.ts test/engine-launch.test.ts test/runtime-config.test.ts test/cli-iii-config.test.ts test/consistency.test.ts
• corepack pnpm run build
• git diff --check
• semgrep scan --config p/default --error --metrics=off .
• gitleaks protect --staged --redact

Local live restart execution was intentionally not run because this workstation already has a user daemon listening on the default iii/REST ports. The new GitHub Actions job is expected to run that live retest on clean hosted runners.

Refs #349, #338, #1033.