modify sync protocol to accept timeout as a parameter

sync persistence and sync replication commands will time out if persistence queue is not drained.
2.5 or 5 or 10 seconds is not going to be enough because we saw that disk write queue size increases to 1M items in amazon deployments and age varies between 0-20 seconds.
things can get worse during rebalancing as well.

by letting the user to specify the timeout we can help them to dynamically increase/decrease the right timeout value based on the cluster state