BUG: write_dataframe to GPKG with an "fid" column gives an error if use_arrow=True

[theroggy]

For some file types it is possible to write "custom" FID's by including an appropriate column in the input dataset, e.g. a column "fid" for GeoPackage.

However, when this is applied using write_dataframe with use_arrow=True this error is thrown: Inconsistent values of FID and field of same name.

An issue for this was opened upstream in GDAL: OSGeo/gdal#11527

EDIT: the conclusions of the upstream issue was that:

1. there was a bug that lead to the error, which is now fixed
2. in pyogrio, the case when there is an "fid" column present isn't handled really well. For formats that explicitly store the FID (e.g. GPKG) columns with the "correct" names (e.g. "fid" for GPKG) are automatically used by GDAL for the FID field. But, in pyogrio, these columns are also added as an ordinary column, which is redundant. The double use of the column is also what triggers the bug in GDAL mentioned in 1., so if this behaviour is changed, the bug isn't triggered anymore: BUG: fix writing fids to e.g. GPKG file with use_arrow #511

Originally noticed in geofileops/geofileops#392