ENH: stats: add random variable infrastructure

[mdhaber]

@tupui rv_infrastructure is the branch you should base your Array API work on. However, I see that there is some work in rv_continuous that was intended to make it into SciPy already - all the stuff in _zeros.py and test_zeros.py. Here's what I'd suggest:

• You can begin to work on _distribution_infrastructure.py and test_continuous.py from rv_continuous.
• Meanwhile, I'll submit a PR to SciPy main with the changes to _zeros.py and test_zeros.py. Sorry those didn't make it before you reviewed the bracketing PR; apparently I had refactored for clarity in this branch. Done.
• After that merges, we'll rebase this branch on SciPy main. We'll also make sure to remove the duplicate hypothesis configuration. Done.
• You might need to merge rv_continuous into your Array API branch again because it will have changed, but there shouldn't be any difficult merge conflicts. You should just be able to use theirs for all files except the ones you've modified.

How does that sound?

[mdhaber]

I think I've responded to all comments. I've resolved some by request and where there were duplicates, typically adding a link back to one that is still open.

I'd like to resolve most of the rest because this page has gotten pretty long, so please go through and resolve the ones you agree have been addressed. If you want to click resolve but GH is not letting you, LMK that I can resolve it using a celebratory emoji (not just thumbs up or eyes, please).

The conversations I think need to be left open are:

• [x] Where to define acronyms
• [ ] Make Sphinx prioritize class over module when resolving links
• [x] "Complementarity" vs "Complement" (now that it's "complement", we're leaving it there)
• [ ] Correct generation of QMC samples

I'd appreciate your thoughts on the first three. The last one is just a task.

[steppi]

@mdhaber, it's not showing me anything I can resolve, but feel free to resolve all of my comments if you can.