replication: Remove 'dead' stale jobs.

When replication job processing is completed we do acknowledgment of the event. If processing was successful the state of the event should be changed to 'completed', for failed processing attempts the state becomes 'failed' and if event processing failed max amount of attempts it should be changed to 'dead'. To reduce amount of rows in the replication_queue table the events that are going to be moved into 'completed' or 'dead' state not updated but removed from the table. The change aligns this behavior to AcknowledgeStale method that was not remove 'dead' events from the table, but only update their status.

Part of: #3859 (closed)