[EPIC] Enable visualization of cloud-hosted geoparquet datasets

[emmalu]

Context

Geoparquet is a relatively new but powerful OGC standard for distributing and visualizing vector data at scale. We can advocate a strong case for why and how all users can leverage this data format, including through Lonboard.

Issue

Enable the visualization of cloud-hosted geoparquet datasets.

Acceptance-Criteria

These are the tasks that need to be completed or artifacts that need to be produced.

• [ ] https://github.com/opengeospatial/geoparquet/pull/191
• [ ] TBD

[kylebarron]

Other possibilities with lonboard specifically are visualizing a cloud-hosted geoparquet dataset. That would only really be possible with the upcoming bounding box definition in GeoParquet 1.1 (https://github.com/opengeospatial/geoparquet/pull/191), so is more of a longer-term goal.

Otherwise much of the tasks here are focused on the JS API, and we may or may not wish to track those inside of the lonboard repo

[emmalu]

Understood, thanks for the corrections @kylebarron. I've updated the issue and left room to expand if/when #191 is complete.

[kylebarron]

I think there's an element of this that may be useful in Python too.

In particular, in geoarrow-rs, we can have a ParquetFile and/or ParquetDataset class for single-file and multi-file parquet data. Then when scanning the metadata (especially with GeoParquet 1.1) we can have a method like "get boxes" where we materialize a table of metadata with each row group and its bounding box. Then we can visualize each partition's extent.

This would be a pretty interesting use case along with some e.g. Overture data, where a user from python could quickly find data in their region of interest.

cc @weiji14 who has been working on remote Parquet file support in https://github.com/geoarrow/geoarrow-rs/pull/493.