Explore ways to improve S2 Geometry performance

S2 Geometry polygon search performance appears to be subpar in several use cases. We need to see if we can leverage

• existing functions and classes in S2
• new research
• new libraries

that can be used to speed up or prune the search space.  https://link.springer.com/article/10.1007/s41019-020-00147-9 seems to imply that S2 can be both very performant and very space efficient on datasets over 300M items.

Examples (to explore - apologies if we are already using some of the below):

• S2ContainsPointQuery
• S2PointIndex to index collections of points in memory and is a variant of Linear Quadtree
• S2ShapeIndex to index arbitrary collections of shapes, i.e., points, polylines and polygons in memory
• S2shape and S2region
• https://s2geometry.io/devguide/s2shapeindex.html 
• ST-Trie: A Novel Indexing Scheme for Efficiently Querying ...
• https://mdpi-res.com/d_attachment/sustainability/sustainability-12-09727/article_deploy/sustainability-12-09727-v3.pdf