ENH: support writing the geoarrow-based encodings of GeoParquet

[jorisvandenbossche]

WIP, on top of #3219

Will probably break out the reading side separately as a first PR, since that does not rely on #3219 (reading side: #3278)

This implements the writing side for geoarrow encodings in GeoParquet (#3253)

[martinfleis]

Looks fine, with my limited GeoArrow knowledge. Some parts of the code are reported as uncovered but at least the ValueError has an explicit test, so I am not sure how trustworthy it is here.

[jorisvandenbossche]

Some parts of the code are reported as uncovered but at least the ValueError has an explicit test, so I am not sure how trustworthy it is here.

Yeah, codecov is doing a bit strange the last days .. Most of the annotations I see are definitely covered

[kylebarron]

How are you thinking of handling the fact that GeoParquet 1.1 hasn't been released yet? Are you thinking of holding this until 1.1 is released?

[jorisvandenbossche]

I was planning to add an "experimental" note to the docstring (it's all opt-in), but yes ideally we would finalize the GeoParquet spec soon (same issue exists for GDAL ..)

[jorisvandenbossche]

(the same issue is true for the bbox writing/reading, so was planning to do that as a follow-up)

We could also add a warning to make it more visible, but given it is purely opt-in a note in the docs might be sufficient?

[kylebarron]

note in the docs might be sufficient

I think that's sufficient

[jorisvandenbossche]

Going to merge this one giving there are a few other PRs that touch the same code