When KafkaConnect is scaled to 0 the reconciliation always timeouts

When the Kafka Connect is set to have 0 replicas, the reconciliation seems to never end and always timeouts. TBH, I was surprised that you can scale it to 0. I think we should:

• Either make it not possible to scale to 0 by better validating the replicas
• Make sure that the reconciliation works fine with 0 replicas.