Revert "[Processor] Add a control message to inform the processor that draining is done"

[TomerShor]

Reverts #3153.

We have seen that this new functionality causes issues and is unstable regarding the different handling of the 2 control message kinds:

• stream ack message is received once by each partition, who always listens for these messages.
• drain done messages are sent from each partition to every worker, meaning that for N partition and M workers - we will have NxM messages. This requires a much more complex logic to handle, which is risky at this point of the development.

Reverting back to only waiting on timeout until we have a better solution to that.

Resolves https://jira.iguazeng.com/browse/NUC-141