Improved site and region assessment processes

[ConnectedSystems]

Avoids relying on any rasters for assessment (although it uses the raster of "valid" pixels to simplify writing results out).

Leverages STR trees (Sort-Tile-Recursive trees) for quick and efficient spatial indexing to reduce search costs compared to indexing rasters or searching dataframes directly.

Also uses Extendable Sparse Matrices to store raster data if the raster size is estimated to be >= 700MB (limit is arbitrary and should be tuned/tunable).

Spatial areas that resolve to be < 700MB use regular matrices and result in faster COG write times (~30% from previous timings).
Sparse matrices reduce memory use by orders of magnitude, but incur additional overhead when writing data out to file.

There's also currently a hardcoded filename suffix ($(region_name)_valid_slopes.tif) - I didn't know where I should put it.
This file is used as a template so that the same extents, spatial coordinates, and CRS are used when creating COGs.

These changes are untested - I don't have the energy tonight to work out how to run ReefGuide methods locally and there's no working example in the readme any more 😢

Note: Most recent processed data includes lons/lats column in the lookup table so do not need to be recreated on data load.
If added a simple check to generate these columns if needed for now.

[PeterBaker0]

Tested - working with very very nice performance improvements, especially for the suitability assessment.

Had some missing variables/errors but have fixed them, tested E2E for regional and suitability with new data. Will need to do a data update on EFS.