Avoid frequent replication restart when node is removed from target cluster

http://qa.sc.couchbase.com/job/cen006-p1-xxdcr-vset07-01-goxdcr-rebalance-ssl/519/consoleText

2017-03-07T23:08:55.616-08:00 INFO GOXDCR.PipelineMgr: Try to fix Pipeline f29a963043bedc655f765304bfc7eb6a/default/default. Current error=xmem_f29a963043bedc655f765304bfc7eb6a/default/default_172.23.105.67:11210_0:Failed to repair connections to target cluster after 5 retries.

We are seeing this in many XDCR jobs (rebalance, bixdcr, lww) - all these jobs end up having rebalance failures - we have MB-23161 filed for the bucket deletion issue - please take a look and see if something needs to fixed on the XDCR side