{DCP/TAP}:: CBTransfer fails {error: could not read _local/vbstate} after Rebalance

1035, Centos 6x

1. Create 3 Node cluster
2. Add a default bucket
3. Load 100 K items
4. Rebalance-in 1 node

Rebalance succeeds and after all the queues have drained, we compare the replica items loaded initially vs present after rebalance-in

The test fails since replica items are missing.

Test Case:: ./testrunner -i ../palm.ini get-cbcollect-info=False,get-logs=False,stop-on-failure=False,get-coredumps=False,force_kill_memached=False,verify_unacked_bytes=True -t rebalance.rebalancein.RebalanceInTests.rebalance_in_after_ops,nodes_in=1,nodes_init=3,replicas=1,items=100000,GROUP=IN;P0

This test case passed for 973 build

https://s3.amazonaws.com/bugdb/jira/MB-11830/replica_keys_missing.txt

https://s3.amazonaws.com/bugdb/jira/MB-11830/1035_log_data_log.tar.gz