feat: Hub monitoring takeover — event push system + config cleanup (v0.21.0)

Replace external Healthchecks.io with Hub-native event system. Controller
now pushes structured events via POST /api/v1/event with typed detail
structs. Hub handles dead man's switch, notification dispatch, and cooldowns.

Phase 5: PushEvent() core method, 21 event types, expanded notification
settings (11 toggles), Hub connection monitoring on dashboard, alerts.
Phase 6: Deprecation log for ping UUIDs, pinger kept for transition.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

controller/internal/web/handler_restore.go

@@ -110,6 +110,18 @@ func (s *Server) executeAllRestores() {
 		return
 	}
 
+	// Count pending apps and push DR start event
+	pendingCount := 0
+	for _, app := range plan.Apps {
+		if app.Status == "pending" {
+			pendingCount++
+		}
+	}
+	if s.notifier != nil {
+		s.notifier.NotifyDRStarted(pendingCount)
+	}
+
+	successCount, failCount := 0, 0
 	for i := range plan.Apps {
 		app := &plan.Apps[i]
 		if app.Status != "pending" {
@@ -126,15 +138,22 @@ func (s *Server) executeAllRestores() {
 		if err != nil {
 			plan.UpdateApp(app.Name, "failed", err.Error())
 			s.logger.Printf("[ERROR] Restore failed for %s: %v", app.Name, err)
+			failCount++
 		} else {
 			plan.UpdateApp(app.Name, "done", "")
 			s.logger.Printf("[INFO] Restore completed for %s", app.Name)
+			successCount++
 		}
 	}
 
 	plan.SetStatus("done")
 	s.logger.Println("[INFO] All app restores completed")
 
+	// Push DR completion event
+	if s.notifier != nil {
+		s.notifier.NotifyDRCompleted(successCount, failCount)
+	}
+
 	// Re-scan stacks so dashboard picks up restored apps
 	if s.stackMgr != nil {
 		if err := s.stackMgr.ScanStacks(); err != nil {