Improve logging in KafkaRoller

There are several issues with logging int the KafkaRoller class:

1. When the pod is assumed to be stuck, it is force rolled. But there is no message logged about this. So it is not clear why the pod was rolled.

1. When the pod is force rolled because the initial connection with the Admin client failed, again, nothing is logged and user does not know why the pod is being rolled.

1. When the pod is rolled because of configuration changes which cannot be done dynamically, it logs this only as a debug message. We should make sure the INFO level in the CO provides explanation why the pod is rolled.

1. When in the same reconciliation there is a change to:

• • to logging configuration which can be done dynamically
  • to broker configuration which cannot be done dynamically
    The restart plan still tries to do reconfiguration before the pod is rolled. This works at the end fine. But it produces this ugly error along the way:

{{2022-03-16 14:50:12 ERROR KafkaRoller:551 - Reconciliation #8(watch) Kafka(myproject/my-cluster): Error doing dynamic config update
org.apache.kafka.common.errors.InvalidRequestException: Invalid config value for resource ConfigResource(type=BROKER, name='1'): Cannot update these configs dynamically: Set(default.replication.factor)}}
This PR fixes the if condition to avoid doing the dynamic update when restart needs to be done anyway.