Support geometries with Z and M values

[Maxxen]

This PR includes a major rework of the runtime geometry representation to enable support for Z and M values. Although no functions that specifically make use of the Z or M axis are implemented yet, and some existing functions may have limited support. Im not terribly happy with the vertex array abstraction yet but it works. Additionally:

There is a new GeometryProcessor abstraction which is used to unify deserialization to GEOS geometries and DuckDB native geometries while also enabling some operations to be performed while deserializing, "zero-copy"-style so as to avoid materializing the geometry hierarchy entirely. The only functions to make use of this so far is ST_Area, ST_AsWKB, ST_(Min/Max)/X/Y/Z/M and ST_AsText.

We now use our own WKT parser instead of relying on GEOS. I don't expect it to be more efficient but it does allow us to handle empty Z and M geometries and provide better error messages (eventually, they're not that exact yet).

New ST_Force2D, ST_Force3DZ, ST_Force3DM and ST_Force4DM functions have been added.

[kylebarron]

Does this support Z and M in the native representations as well or just in the GEOS one?

[Maxxen]

Native too! + WKB/GeoJSON/GDAL support. Although I haven't added new vectorized representations like POLYGON_2D

[kylebarron]

Although I haven't added new vectorized representations like POLYGON_2D

ah that was what I meant to ask!