[FTS] MCP should perhaps pause ingest for pindexes on nodes that are rebalancing out as an optimization?

Found this when diagnosing MB-18385...

When you first create an index, many pindexes will be assigned across the various cbft nodes. And, those cbft nodes will dutifully begin their DCP streams to start building up those brand new pindexes.

Then, when you next kickoff a rebalance, the MCP does not stop the pindex building, even on nodes that will be soon going away. That is, there's a whole bunch of wasteful looking activity still going on.

To do that, probably is a bigger thing that can fit into beta.

Or, for now, a sufficiently knowledgable user can also use the "secret" UI to perhaps "pause" index ingest before hitting the rebalance button.