Optimization for multi-GEOMETRY on query and ingest

[chrisbennight]

When multipolygons are used currently for query or when indexed the MBR of the entire feature is used. It would be beneficial to instead operate each individual feature in the multi-* feature, and

• in the case of query the set of ranges would be the set of ranges decomposed for each individual feature
• in the case of indexing the set of indexed MBR's should be the set of MBR's for each individual feature (polygon, etc.). There's also a decision here if we should exempt this from our "cap" on feature decomposition or not. This may exempt multipolygons for the decompsed region cap (or, more correctly, set the decomposed region cap at max regions * # of sub features.

[chrisbennight]

Discuss: if we pre-split the multi-geometry into multiple geometries and send that in to ingest, we could end up with a case where a given feature is stored on multiple tiers.