fix(desktop): a primary backend crash loop now shows why it can't start

The primary backend could exit before readiness on every restart (for example
with a corrupt statev2.sqlite) and the user saw no window and no error, because
the renderer is served by that backend.

After three consecutive exits without readiness, the pool now shows a native
error box with the exit reason and the server-child.log path. Restarts keep
going in the background, and the streak resets once the backend becomes ready.

Closes #10517

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ad92ed834bJulius Marminge committed on 10/22/2026, 8:46:06 PMparent62b3961
3 files changedLine totals unavailable