Seems really odd that those would be different.

I just pushed an update that puts the same logic into each of the readers so there isn't any difference between FionaReader and BasicReader in how you get a WKT and we rely on pyproj's version to decode the .prj file if it is present.

Not sure what's up with CI.

I'm going to move this back to draft for now so it doesn't get merged.

The docs failures are real and I assume due to an interaction of this new CRS for NaturalEarth and the handling updated in #2455
I can reproduce it with a local docs build, but just running the examples manually as a script works fine. I'm not sure what specific figure size and other options sphinx_gallery uses, but my guess is there is some different settings in there that cause the error to show up.

The other thing I noticed looking through the logs is that we have a lot of UserWarnings now:

UserWarning: Approximating coordinate system <Geographic 2D CRS: GEOGCRS["WGS 84",DATUM["World Geodetic System 1984 ...>
Name: WGS 84
Axis Info [ellipsoidal]:
- lon[east]: Longitude (Degree)
- lat[north]: Latitude (Degree)
Area of Use:
- undefined
Datum: World Geodetic System 1984
- Ellipsoid: WGS 84
- Prime Meridian: Greenwich
 with the PlateCarree projection.

which means that UserWarning gets printed a lot because it is associated with every natural earth feature now.

OK, I got back to looking at this again and will put it up for review. I added a new commit that fixes the previous issues I was running into.

• Changing to a ccrs.Projection() instead of base ccrs.CRS() provides a way to get the bounds and avoid all of the warnings that we are converting to a Projection instance later.
• Setting default bounds of global for geographic CRS' because Natural Earth .prj files don't specify an area of use, but we can default to a geographic (lon, lat) coordinate system having a global domain.