feat(python/sedonadb): Support specifying layer name and accessing sub-paths inside archives when working with pyogrio sources.

[pratheekrebala]

Summary

• Enables passing pyogrio keyword arguments (e.g., layer, where, sql, max_features) through read_pyogrio(options=...) by switching from the internal pyogrio.raw.ogr_open_arrow() to the public pyogrio.raw.open_arrow() API, which accepts these as keyword arguments directly including specifying the layer= flag to handle multi-layer files (python/sedonadb: No mechanism to pass layer parameter to read_pyogrio #772)

• Adds a path_suffix option to append a subpath to the resolved GDAL source, enabling reading of formats like FileGeoDatabase (.gdb) stored inside .zip archives (e.g., options={"path_suffix": 'subdir/test.gdb')). This is equivalent to running ogrinfo /vsizip/test.zip/subdir/test.gdb.

Motivation

Previously, read_pyogrio() had no way to pass options through to pyogrio, making it impossible to select a specific layer in a multi-layer file or use GDAL driver-specific open options. For example, reading a named layer from a GeoPackage now works:

con.read_pyogrio("file.gpkg", options={"layer": "my_layer"})

For formats like FileGeoDatabase distributed inside .zip archives, GDAL expects a path of the form /vsizip/archive.zip/data.gdb. Because the file listing infrastructure resolves real filesystem paths before the Python reader runs, passing a full /vsizip/... path directly fails. The path_suffix option bridges this gap:

con.read_pyogrio("path/to/file.zip", options={"path_suffix": "data.gdb"})

[paleolimbot]

Thank you!

In CI I see:

        # column comparison
>       assert_index_equal(
            left.columns, right.columns, exact=check_column_type, obj="GeoDataFrame.columns"
        )
E       AssertionError: GeoDataFrame.columns are different
E       
E       GeoDataFrame.columns values are different (50.0 %)
E       [left]:  Index(['val', 'geom'], dtype='str')
E       [right]: Index(['val', 'wkb_geometry'], dtype='str')

and left a few suggestions for how to fix the test. I'm never quite sure what the name of a geometry column will be when I write or read a GeoDataFrame (everybody chooses something different for the "no geometry column name" case).

[paleolimbot]

Thank you!