I do not understand the need to have access to the polygons. The cache-data already has the counties assigned. I must be missing a point here.
Regarding how the GenericPolyChecker does it, is that lillfiluren has provided us with JSON data, that we store server side (well, the scripts are on the servers as well). This to avoid having 50 MB of data in the script, which makes the editor and saving slow. We can do this for anyone, it's just a way to store data that is needed for the script.
Regarding mole125's note. Having access to the polygon data itself isn't always practical. It can be many hundreds of megabytes for a country. We use as detailed polygons as we can. Here as well I would need a use case to actually understand the need.
We do not have any contintal polygons either, more or less not even for countries. We mostly have for regions and counties, which is what we have needed.
PS! We have on our list to upgrade the polygons for France when time allows. It's a bit tricky though.