Improve index rebalance behavior when index build is in progress

The behavior was improved in MH from "unconditional fail" to "check if an index build is required as part of rebalance on the node where another index build is already in progress".

The behavior is still less than desirable as a rebalance failure is disruptive for other services as ns-server aborts the rebalance operation completely. 

Few possible improvement options:

1. Abort the running index build to allow rebalance to proceed. Let the builder rebuild the aborted build after rebalance is done.
2. Rebalancer can examine the planner output and determine if any essential index movement is required. If not, index rebalance could be marked success without doing anything.
   1. Essential index movement - Index(es) on a node moving out of the cluster. 
   2. Non-essential index movement - Moving around partitions of a partitioned index for balanced resource utilization.
   3. Replica repair needs more thought.