Initial Sync is aborted when the task is rebalanced

Context

Initial Sync is aborted when the task is rebalanced, it's normal that some initial syncs might take longer, hours and days even.

When KC decides to rebalance a task by its own internal state, when a worker has died or any external influence in the healthiness of the cluster, the task will stop without considering there's an initial state ongoing.

Suggestion

Wait for the initial sync to finish before allowing a rebalance

Implementation

Improve the `stop()` callable to wait for an initial sync to be done
I'm not sure how much we can push the boundaries here, but if it's enough to avoid issues for rebalances caused by added workload (like when you add a new connector) that should prevent majority of the cases.