QueueShutDown subclass doesn't catch native asyncio.QueueShutDown on Python 3.13+ — spurious error logged on every shutdown

[FarhanAliRaza]

Problem

On Python 3.13+, every EventProcessor.stop() logs

Error in event processor queue task during shutdown:
...
asyncio.queues.QueueShutDown

and sends the exception to telemetry, even though shutdown is working as intended.

Mechanism

(packages/reflex-base/src/reflex_base/event/processor/event_processor.py)

The compat shim defines a subclass of the native exception:

if hasattr(asyncio, "QueueShutDown"):
    class QueueShutDown(asyncio.QueueShutDown):
        ...
else:
    class QueueShutDown(Exception):
        ...

Instance checks only work one way: the local class catches itself, but the native asyncio.QueueShutDown raised by queue.get() after stop() calls self._queue.shutdown() is not an instance of the local subclass. As a result:

• with contextlib.suppress(QueueShutDown): around the _process_queue loop does not suppress it, so the queue task dies with the native exception, and
• in stop(), except (asyncio.CancelledError, QueueShutDown, RuntimeError): when awaiting self._queue_task doesn't match either, so the native exception falls through to except Exception, which calls telemetry.send_error(...) and console.error("Error in event processor queue task during shutdown: ...").

Pre-3.13 is unaffected (the queue is never shut down and the local class is only raised, never caught-against-native).

Fix sketch

Alias instead of subclassing when the native exception exists:

if hasattr(asyncio, "QueueShutDown"):
    QueueShutDown = asyncio.QueueShutDown
else:
    class QueueShutDown(Exception):
        ...

(or catch asyncio.QueueShutDown explicitly alongside the shim at both sites).

Found while adding drain assertions to the event-loop benchmarks in #6751.

[linear-code]

ENG-10196