I think it's something unique to python's ML ecosystem, to be honest. There is a lot of up-in-the-air about how to handle models, binaries and all of that in a contained package, and that results in quite a few hand-rolled solutions, some of which encroach on the package manager's territory, plus of course drivers and windows.

I've worked on/with some fairly large/complex python projects, and they almost never have any packaging issues that aren't just obvious errors by users. Yes, every once in a while we have to be explicit about a dependency because some dependent project isn't very strict with their versioning policy and their API layers.