Optimize the CRUS features collection

[doublebyte1]

Original layer (for the whole country): https://snig.dgterritorio.gov.pt/rndg/srv/por/catalog.search#/metadata/517c5023-04cc-47a4-99f7-bb32814dd62f

pygeoapi config: https://github.com/doublebyte1/ogcapi-simple/blob/master/pygeoapi/docker.config.yml#L136

Some ideas:

• on the storage: Optimize the DB table with indexing
• on pygeoapi: restrict the number of features that can be served by default

[doublebyte1]

The upgrade to a newer version of PostGIS reduced the loading time of vector tiles to 50%: https://github.com/byteroad/ogcapi-simple/commit/5c285d5fa1be44179f7db858176e0b960334dcb2

[doublebyte1]

Tests:

• [x] serve MBTiles from file
• [ ] tune martin's config.yml
• [ ] tune alchemy's connection
• [x] Render tiles on disk with tippecanoe