Previously, the reconnection loop was calling client.disconnect() during cleanup,
which triggered the disconnect callback with code 0 ("Normal disconnection").
The callback would then exit early, preventing further reconnection attempts.
This creates a cleanup flag that prevents the disconnect callback from
stopping reconnection when we're intentionally cleaning up the old client
during retry attempts.
Fixes issue where MQTT would get stuck in retry loop without actually
attempting fresh connections.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add _reason_info helper method to extract reason names safely
- Update connect/disconnect callbacks to use string comparison instead of numeric
- Fix type annotations with proper Optional[Client] typing
- Resolve unreachable code warnings with type ignore comments for threading
- Improve error handling with robust try/finally blocks in stop method
- Maintain fresh client creation approach for reliable reconnection
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Replace client.reconnect() with fresh client creation for each retry attempt
- Add proper cleanup of old client before creating new one
- Enhanced logging with detailed debugging info for disconnect reasons
- Use proven aiomqtt-style retry pattern for better reliability
- Each reconnection attempt now creates completely new MQTT client instance
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
