Rebalancer should utilize full transferBatchSize for index movements

With the smart batching algorithm introduced in 7.x, rebalance divides the transferBatchSize quota equally among all indexer nodes. In case, there are not enough index movements for all the nodes(e.g. replica repair on a single node), the batch size can be under utilized. This can lead to slower rebalance as well as it is not intuitive for the customer.

It is better to improve this such that rebalancer utilizes full transferBatchSize for index movements.