chore: dev-stack switches, shared e2e harness, and desktop shell rules

- `LIVE_RELOAD=false npm run dev` keeps the renderer alive across a machine
  suspend; the reload client otherwise destroys the session under test.
- `dev-peer.sh` plus a separate userdata dir runs a second local peer.
- `tools/voice-probe.js` samples peer state and RTP counters from a live
  window, persisting to localStorage so a renderer reload cannot erase it.
- e2e helpers for voice pairs, peer-role election, and a TURN relay.
- Electron single-instance and dev-client-load decisions move into rules
  files with colocated specs.

emergency-fix/e2e-failures/slow-tests.md

@@ -0,0 +1,17 @@
+# Slow E2E tests (≥30s)
+
+All of these **passed**. Listed because the product ask is that timeouts should not happen and the app should stay fast.
+
+| Duration | File | Test |
+| --- | --- | --- |
+| 128.2s | `voice/mixed-signal-config-voice.spec.ts` | 8 users with different signal configs can voice, mute, deafen, and chat concurrently |
+| 125.1s | `voice/multi-signal-eight-user-voice.spec.ts` | keeps 8 users on 2 signal apis while voice, mute, and deafen stay consistent for 20+ seconds |
+| 53.4s | `auth/user-session-data-isolation.spec.ts` | gives a new user a blank slate and restores only that user local data after account switches |
+| 33.7s | `chat/chat-message-features.spec.ts` | shows per-server channel lists on first saved-server click |
+| 30.3s | `chat/multi-client-chat-sync.spec.ts` | syncs messages between same-user devices and late-joining users after offline gaps |
+
+## Notes
+
+- `mixed-signal-config-voice` and `multi-signal-eight-user-voice` are intentionally heavy (8 users / multi-signal). ~2 minutes each is suite design cost more than a single-user app hang.
+- `user-session-data-isolation` (~53s) and channel/chat sync (~30–34s) are closer to “app feels slow” candidates if they regress upward.
+- Playwright config `timeout` is 90s per test; none of the failures/timeouts hit that ceiling in this run.