[Feature] Local Play should warn when the local dev server is down instead of failing silently

[ranjith7022]

Summary

When the Vite dev server stops while a code app is open through the Local Play URL, the app keeps running normally with no visual indication that anything changed. HMR is silently dead — edits no longer appear — so developers often don't notice until their changes "don't work." Refreshing during the outage shows only "We could not retrieve your app."

This is a common workflow interruption (dev server crash, accidental Ctrl+C, restart onto a different port that stales the saved Local Play URL), and today there's no out-of-the-box signal at any layer.

Environment

• @microsoft/power-apps-vite 1.0.2
• Vite 8.0.13, @microsoft/power-apps-cli 1.0.0 (pa app run)
• Windows 11, Chrome 153

Repro steps

1. Run pa app run, open the printed Local Play URL
2. Kill the dev server process
3. Observe: the loaded app continues working normally — no error, no warning
4. Edit any source file → nothing happens (HMR websocket is gone)
5. Refresh while the server is still down → player error screen

What would help

Any of the following would address it — happy with whatever the team thinks fits best in the platform:

1. In-app indicator injected by the Vite plugin — e.g. a banner/toast when the HMR websocket drops, cleared on reconnect. I validated this approach locally using Vite's client events (vite:ws:disconnect / vite:ws:connect) and it works inside the Local Play iframe.
2. Player-side improvement — when the player can't reach _localAppUrl/_localConnectionUrl, show an actionable message ("Your local dev server at http://localhost:5173 isn't reachable — restart pa app run") instead of the generic "We could not retrieve your app."
3. CLI-level guardrails — e.g. a stable/port-pinning option for pa app run so saved Local Play URLs don't go stale after a restart.
4. Docs — if none of the above are planned short-term, a note in the debugging docs about this failure mode and its recovery path.

Why it matters

• The embedded-player workflow makes the existing console-only message ([vite] server connection lost. Polling for restart...) effectively invisible
• The current refresh error doesn't point at the actual cause, so developers debug their app instead of their terminal