Adaptive Indexes and (Sub) Index Distribution

Adaptive indexes are a great idea however are a single node resource hog for large datasets. Need to vertically scale out but we don't have the options today to scale out a single index node vertically - see MB-23177

In the meanwhile would be awesome if we could auto-distribute the adaptive index across multiple nodes to reduce the per node capacity requirement.

e.g. with NODES { node 1, node 2, node 3} should distribute portions of the index across the 3 nodes.

Obviously answer depends upon whether we are implementing auto index sharding in 6.0

if yes, then this would not be required

if not, we need to have the ability to distribute the sub-indexes across multiple node