fix: prevent agents from running indefinitely after app quit

Agents kept running after closing AgentCompany because worker processes
auto-recreated on exit, retry timers were never cleared, child CLI
processes weren't terminated, and orchestration had no quit guards.

Changes:
- Add isQuitting guards to all orchestration/automation entry points
- Track and clear all setTimeout/setInterval timers on quit
- Send cancel-run to all active CLI processes before killing worker
- Add destroyed flag to prevent worker recreation after intentional destroy
- Add quit confirmation dialog: Stop All & Quit / Minimize to Background / Cancel
- Bump version to 0.4.96

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

src/main/bootstrap-application.ts

@@ -192,6 +192,7 @@ export async function bootstrapApplication({
   // Re-register timers for scheduled social posts (survives app restart).
   // Scan all companies for queued browser actions with a future scheduledAt time
   // and set up setTimeout to dispatch them at the right moment.
+  const scheduledPostTimers = new Set<ReturnType<typeof setTimeout>>();
   try {
     const snapshot = db.listSnapshot();
     for (const company of snapshot.companies) {
@@ -203,12 +204,15 @@ export async function bootstrapApplication({
             if (payload.scheduledAt) {
               const delay = new Date(payload.scheduledAt).getTime() - Date.now();
               if (delay > 0) {
-                setTimeout(() => {
-                  const browserActions = serviceGraph.getBrowserActionService();
+                const timer = setTimeout(() => {
+                  scheduledPostTimers.delete(timer);
+                  if (isQuittingRef.current) return;
+                  const browserActions = serviceGraph?.getBrowserActionService();
                   if (browserActions) {
                     void browserActions.dispatchBrowserAction(action.id, company.id);
                   }
                 }, delay);
+                scheduledPostTimers.add(timer);
                 log.info(`[bootstrap] Re-registered scheduled post timer: action ${action.id} in ${Math.round(delay / 1000)}s`);
               }
               // If delay <= 0, the action is overdue -- normal dispatch cycle will pick it up
@@ -221,6 +225,16 @@ export async function bootstrapApplication({
     log.warn(`[bootstrap] Failed to re-register scheduled post timers: ${error instanceof Error ? error.message : String(error)}`);
   }
 
+  // Register cleanup for bootstrap-owned resources during quit.
+  // This runs BEFORE service-graph's onBeforeQuit (which is registered later).
+  app.on("before-quit", () => {
+    for (const timer of scheduledPostTimers) {
+      clearTimeout(timer);
+    }
+    scheduledPostTimers.clear();
+    notificationService.destroy();
+  });
+
   // Show the window immediately so the user sees the app launch without delay.
   await serviceGraph.getDesktopShellService()?.ensureWindow();
   serviceGraph.getAppLifecycleService()?.register(hasSingleInstanceLock);