This is not a perfect workaround but it seems to be the best solution
for the moment.
On Windows, this means that when a patch fails to apply, you would now
need to fix that patch, finish the apply (AM), then rebuild all patches,
and then finally re-run the patch apply procedure in order to continue.
This adds a small amount of overhead compared to the traditional method
(which will still work on *nix environments, including WSL). However, it
seems preferable to the build not working on Windows at all.