Improvement: built-in facility for measuring XDCR latency

This improvement is a request for a heartbeat facility/API for each XDCR stream replication so any program can easily check the replication lag to a target. The customer could then monitor this in real-time as part of their operational monitoring system.

Right now, there is no such feature built-in and would require customer to manually set up something for each bucket being replicated.