I wrote up some simple decision tables for the kata, and used them to write a thorough set of tests. From there, I was able to rewrite the kata code with a much flatter control flow. Passed all tests once the types lined up properly.
It's late now, so I'll work on the next part tomorrow.
Ward was mostly easy to use, although I initially expected parameterization to work a little more like in pytest, which it doesn't quite. Basically, Ward's parameters cannot vary independently. This required me to split up some tests a little more thoroughly, which is probably fine.
I guess I don't have much to say, since putting together the tables then translating them to tests was pretty straightforward and quick. No real complaints; I'd definitely recommend at least trying it for any given project.