Fix Docker restart crash: respect autoStart flag and persist identity

Root cause: on Docker restart, loadSavedBots() unconditionally connected
ALL saved bots regardless of autoStart flag, causing a rapid
connect/disconnect loop. Additionally, bot identities were regenerated
on every restart, causing TS server conflicts with stale sessions.

Changes:
- loadSavedBots() now only auto-connects bots with autoStart=true
- startBot/stopBot persist autoStart state so restart behavior matches
  user intent
- Bot TS3 identity is persisted to database and reused across restarts
- Database schema migrated to include identity column
- TS3Client.connect() cleans up existing connection before reconnecting
- Stagger bot connections by 1s to avoid overwhelming the TS server

https://claude.ai/code/session_01L2kEV2M1QFWMCyPtLU5LgC

src/ts-protocol/client.ts

@@ -53,6 +53,18 @@ export class TS3Client extends EventEmitter {
   }
 
   async connect(): Promise<void> {
+    // Clean up any existing connection before creating a new one
+    if (this.client) {
+      this.logger.info("Cleaning up previous connection before reconnecting");
+      try {
+        await this.client.disconnect();
+      } catch {
+        // Ignore errors during cleanup
+      }
+      this.client = null;
+      this.clientId = 0;
+    }
+
     const addr = `${this.options.host}:${this.options.port}`;
     this.logger.info({ addr }, "Connecting to TeamSpeak server (full client protocol)");