Provide more performant storing options when importing geo data from maps

[iverase]

Kibana maps currently supports importing geo data from geojson and shape files. In both cases it stores the geometry using a standard geo_shape field with the geometry encoded as geojson. This is probably not the most efficient way to store a geometry when it is going to be consumed by the vector tiles API so I propose to give some additional options on how to store them:

• In Elasticsearch 8.8, it is possible to store a geometry as a store field (using store parameter set to true). The geometry is stored as WKB and the vector tiles will use that field if present as it is much faster to read the geometry from there. I wonder if we should give the option to the user to store the field, this will trade read performance from disk space.

• Maybe when using a shape file, we can give the user the option to store the geometry as WKT. This format is more efficient to read from source than geojson. This is because when er read from source, we create an object model of the document in memory. In the case of geojson this means a lot of small allocations which is not GC friendly. In the case of WKT it is just allocating a string which is better when reading several documents.

[elasticmachine]

Pinging @elastic/kibana-presentation (Team:Presentation)

[ThomThomson]

Closing this and adding it to the maps icebox