The topic operator's liveness/health checks currently just return 200 to all queries. However, if the TO is using the KafkaStreams based state store and it has entered an unhealthy state, the TO is no longer functional, and the liveness probe should reflect that.

Acceptance criteria

• When TO is using KafkaStreams state storage, it's ready when the streams instance is in a RUNNING state
• When TO is using KafkaStreams state storage, it's live when the streams instance is in any state other than NOT_RUNNING or ERROR.

Please see the comments in this code for descriptions of the possible states and transitions.