Consumer.stop() blocks in GroupCoordinator.ensure_coordinator_known() if all brokers are unavailable

[dkilgore90]

Describe the bug
In a scenario where all kafka brokers have been shut down, an application may also want to shut down (or at least stop consumers). Consumer.stop() can get blocked, cycling endlessly to confirm the group coordinator id:

https://github.com/aio-libs/aiokafka/blob/v0.6.0/aiokafka/consumer/group_coordinator.py#L535-L541

the NodeNotReadyError received in this case is marked as retriable, so the loop continues.

Expected behaviour
Consumer.stop() should be able to complete

Environment (please complete the following information):

• aiokafka version (python -c "import aiokafka; print(aiokafka.__version__)"): 0.6.0
• kafka-python version (python -c "import kafka; print(kafka.__version__)"): 2.0.1
• Kafka Broker version (kafka-topics.sh --version): 2.3.0
• Other information (Confluent Cloud version, etc.):

Reproducible example

# Add a short Python script or Docker configuration that can reproduce the issue.