FTS - Query supervisor duplicate the query accounting on every node

The slow query monitoring feature is tracking the search queries on every node with different IDs specific to that node.

This won't help in managing the query at a cluster level.

We can better this by tracking it only on the query coordinating node. 

A UI could always hit the endpoint on all the cluster nodes and aggregate the results in the UI.

It could also track the associated server node with each result now (under display) so that it

can issue the cancel request to the correct coordinating node.