fix: medium severity audit fixes — races, leaks, perf, warnings

- Clear teams/teamTasks maps in handleInit() on SSE reconnect
- Fix saveNowAsync() race condition with in-flight promise guard
- Add setMaxListeners() on Session, WebServer, SubagentWatcher,
  ImageWatcher, TmuxManager to prevent MaxListenersExceeded warnings
- Debounce persistSessionState() per-session (100ms) to reduce
  redundant toState() serialization across 25+ call sites
- Cache getLightSessionsState() with 1s TTL to avoid re-serializing
  all sessions on every SSE connect and /api/sessions request

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

src/session.ts

@@ -369,6 +369,7 @@ export class Session extends EventEmitter {
     niceConfig?: NiceConfig;  // Nice prioritying configuration
   }) {
     super();
+    this.setMaxListeners(25);
 
     // Default error handler prevents unhandled 'error' events from crashing the process.
     // Server attaches its own handler after construction — this is a safety net for the gap.