Structured Release Process

[wdm0006]

Historically this project has done releases based on when I felt like doing a release, or when someone explicitly opened an issue asking for one. That's not great.

I'd like to discuss how this community would like to do releases (scheduled? based on key features/bug fixes?), codify that in a markdown file in the repo, and then support that via the CI and defined processes, roles, and responsibilities.

To start with, I'll propose the simplest option I can think of: monthly releases each following semver versioning. If there are no commits in a given month, we skip it. Currently the project is hosted under my personal PyPI acct, I'm not sure how other scikit-learn-contrib projects are handling that, but we may should change that as well.

Thoughts?