Stabilize draft migration E2E relaunch seeding

[sambitcreate]

Summary

• add a post-shutdown/pre-launch hook to the Electron E2E relaunch helper
• seed migration fixtures only after Electron has fully closed
• cover all draft-migration tests that previously edited persisted files while the app was running

Why

Main CI for the 0.41.5 release twice classified the draft migration test as flaky because Electron shutdown could flush an older in-memory chat index over the test fixture. This is test-harness-only; the signed/notarized 0.41.5 release workflow passed and production code is unchanged.

Verification

• npm run type-check:e2e
• npm run build
• focused draft-chat E2E with --repeat-each=3 --fail-on-flaky-tests (15 passed)
• npm run type-check
• npm run lint
• git diff --check

[pullfrog]

✅ No new issues found.

Reviewed changes — Reviewed the post-shutdown relaunch hook, migration fixture updates, and troubleshooting note.

• Relaunch lifecycle — The fixture closes the prior Electron process before running the optional async callback and launches the replacement only after all callback writes complete.
• Migration coverage — The draft-chat scenarios now seed persisted chats, journals, receipts, and markers only between relaunches, preventing shutdown persistence from overwriting the fixtures.
• Troubleshooting record — The papercut entry documents the shutdown-seeding failure mode for future E2E work.

  ^{｜ View workflow run ｜ Using GPT Luna ｜ 𝕏}