Type annotations

[einarwar]

Hi there!

I am working on a project where we use Monte Carlo sampling, and we originally used functions from semeio for that. However, it seems like probabilit is a much better fit for our use case, since it only contains the parts we need. Also with the recent semeio releases, we took the hint that we should use this package instead :)

I've been bothering @berland with adding some type-annotations over there, and observed that probabilit also could use some annotations. Do you have any thoughts regarding this? I'm more than happy to provide assistance if needed

[tommyod]

I personally see little advantage of types in code bases like these (scientific, small, only 2-3 user facing functions (.sample() method, Correlators, nearest_correlation_matrix)). Of course there will be some advantage and I am sure we'll find some type related bug(s) if we added types, but the cost in terms of writing out types, maintenance of types, false positive errors, extra linting work on every PR, etc might not be worth it. Taking opportunity cost into account.

That being said I would also welcome a PR. But then please start small - add a few types to where you really think it provides a benefit, then we can have a discussion.