fix(PROJ-86): stale sync_running-Flag beim Scheduler-Start automatisch bereinigen

Ein IMAP-Konto blieb wegen sync_running=true seit 2026-07-05 (Prozess-Crash
vor dem defer in runSyncWithRetry) für den Scheduler auf ewig "beschäftigt" -
checkAccounts überspringt Konten mit SyncRunning=true. Neue Store-Methode
ClearStaleSyncRunning() wird jetzt bei jedem Scheduler.Start() aufgerufen und
setzt verwaiste Flags zurück, bevor der Scheduler-Loop startet.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UPFC6Jk2ke1Pq9XcuVGP1R

internal/imap/store.go

@@ -349,6 +349,21 @@ func (s *Store) UpdateSyncInterval(ctx context.Context, id int64, intervalMin in
 	return nil
 }
 
+// ClearStaleSyncRunning resets sync_running=true for every account back to
+// false. Called once at process startup: sync_running is only ever set on
+// this same process, so any row still marked running is a leftover from a
+// prior process that crashed/was killed mid-sync (the defer in
+// runSyncWithRetry never ran) — without this, the scheduler would skip that
+// account forever (checkAccounts skips accounts with SyncRunning=true).
+// Returns the number of accounts reset.
+func (s *Store) ClearStaleSyncRunning(ctx context.Context) (int, error) {
+	tag, err := s.pool.Exec(ctx, `UPDATE imap_accounts SET sync_running = false WHERE sync_running = true`)
+	if err != nil {
+		return 0, fmt.Errorf("imap store: clear stale sync running: %w", err)
+	}
+	return int(tag.RowsAffected()), nil
+}
+
 // SetSyncRunning marks whether a background sync is currently active for an account.
 func (s *Store) SetSyncRunning(ctx context.Context, id int64, running bool) error {
 	_, err := s.pool.Exec(ctx,