The Firestore watch/listener system tests (test_watch_document, test_watch_collection, test_watch_query, and test_watch_query_order) currently rely on hardcoded sleep(0.2) delays and polling loops to coordinate snapshot updates.

These delays are fragile, highly susceptible to environmental latency/slow networks, and can cause test flakiness. We should investigate replacing these arbitrary polling loops with a robust, event-driven synchronization mechanism (e.g., using threading.Event or callback signals) that yields immediately upon receiving the expected snaps.