fix(#1356): Replace HTTPServer with ThreadingHTTPServer for concurrent users

The multi-user bridge used single-threaded HTTPServer which processed
requests sequentially. Each LLM inference call (5-10s) blocked all
other users, causing 60% timeout rate at 10 concurrent connections.

Changes:
  - world/multi_user_bridge.py: Added ThreadingMixIn import,
    ThreadingHTTPServer class with daemon_threads, swapped
    HTTPServer -> ThreadingHTTPServer in main()
  - multi_user_bridge.py (root): Fixed main() to use the already-
    defined ThreadingHTTPServer class instead of raw HTTPServer

Fixes #1356

world/multi_user_bridge.py

@@ -34,9 +34,10 @@ from typing import Optional
 
 
 class ThreadingHTTPServer(ThreadingMixIn, HTTPServer):
-    """Thread-per-request HTTP server."""
+    """Thread-per-request server for concurrent multi-user handling."""
     daemon_threads = True
 
+
 # ── Configuration ──────────────────────────────────────────────────────
 
 BRIDGE_PORT = int(os.environ.get('TIMMY_BRIDGE_PORT', 4004))