Refactor DatabricksSqlHook timeout handling to use explicit signaling

[SameerMesiah97]

Description

This change refactors timeout handling in DatabricksSqlHook.run() to replace implicit timeout detection based on Timer.is_alive() with explicit signaling using a threading.Event. Previously, timeout classification relied on inferring whether the background timer thread had completed, which coupled error handling logic to thread lifecycle behavior. This refactor introduces explicit timeout signaling in create_timeout_thread() and updates run() to check that signal instead of inspecting thread state.

The refactor preserves existing cancellation semantics and exception types while removing reliance on thread liveness inference.

Rationale

The previous implementation inferred timeout occurrence from the lifecycle state of a threading.Timer, making the logic harder to reason about and tests dependent on implicit thread behavior. The inline TODO noted that this logic should be adjusted to make testing easier. By introducing explicit timeout signaling, timeout classification becomes deterministic, clearer, and easier to test without relying on thread internals.

Tests

Timeout-related tests were updated to simulate timeout conditions by mocking create_timeout_thread and the associated timeout signal rather than relying on threading.Timer behavior.

Backwards Compatibility

No changes to public APIs or behavior.

[SameerMesiah97]

Requesting review for this.