OOI, what possibilities? I'm genuinely interested know what problems PEP 622 would solve for you.
It's somehow the same use case as Oscar Benjamin's: transform algebraic expressions (the goal being to calculate them step by step -- still under development). You have a series of rules to apply, like transforming x*1 into x, and more complicated ones. These rules can be written naturally using patterns.
Also, I have the loose project of writing a new ABC to LilyPond converter (both are text-based music notation formats). This is all about manipulating different tree structures held in dataclasses, so the match statement, possibly in conjunction with algebraic types, would be useful as well.
Best, Jean Abou Samra