Add GenGammaRV jax implementation

[FredericWantiez]

Adding a jax implementation of GenGammaRV, closing #1333. Could do better with the test I've setup, but I don't think stats.cramervonmises can take a named scale argument.

Here are a few important guidelines and requirements to check before your PR can be merged:

• There is an informative high-level description of the changes.
• The description and/or commit message(s) references the relevant GitHub issue(s).
• pre-commit is installed and set up.
• The commit messages follow these guidelines.
• The commits correspond to relevant logical changes, and there are no commits that fix changes introduced by other commits in the same branch/BR.
• There are tests covering the changes introduced in the PR.

Don't worry, your PR doesn't need to be in perfect order to submit it. As development progresses and/or reviewers request changes, you can always rewrite the history of your feature/PR branches.

If your PR is an ongoing effort and you would like to involve us in the process, simply make it a draft PR.

[brandonwillard]

Looks like this is failing in CI because the Python 3.7 tests are using SciPy 1.61 and not > 1.7.0.

[brandonwillard]

We can probably change the SciPy lower bound to 1.7.0, because our NumPy lower bound is 1.17, and SciPy recommends a version greater than 1.7.0 for that. The only relevant SciPy constraint of which I'm aware is in numba-scipy, which currently upper-bounds SciPy at <= 1.7.3, but allows the desired 1.7.0 lower bound.

[rlouf]

I just tried and the tests pass locally with SciPy==1.7.2.

[rlouf]

The alternative is to add the test to test_random_RandomVariable which uses stats.cramervonmises.

[FredericWantiez]

I didn't use stats.cramervonmises in test_random_RandomVariable as it doesn't really allow you to handle named arguments and scipy also uses a different parametrization. I'm happy to change if bumping scipy just for this is not an option

[rlouf]

You can use a different parametrization in the tests, look at the test for the gamma distribution for instance. Unless I misunderstood your problem?

[FredericWantiez]

Nice, somehow missed that. I've pushed something to use test_random_RandomVariable but feels brittle to force the loc and scale arguments this way. Some details on the issue with cramervonmisses here: scipy/scipy#18055

[brandonwillard]

FYI: I've rebased, squashed, and fixed the formatting issue from the last CI run.

[FredericWantiez]

Sorry for the spam, rebased the wrong branch. Let me quickly fix it

[brandonwillard]

I've rebased, adjusted the docstring, and increased the number of samples drawn for the Cramér–von Mises test. This should be good to merge after the tests pass.