Matt Haberland

This looked close, so I went ahead and `export`ed only float32 and float64, and I use `np.result_dtype` to determine which of those dtypes to convert to before calling the function....

This PR renames `_hypotests_pythran` to `_stats_pythran` (since it's not just for hypothesis tests) and copy-pastes (with mild modifications) the innards of `siegelslopes` into `_stats_pythran` for a nice speed boost It...

There is a similar failure showing up in linux_aarch64_test that was not xfailed by https://github.com/scipy/scipy/pull/17057. I opened gh-17273 to xfail that.

> Benchmarking should have been a requirement for the latest addition, propack to be merged, but it was not. It was. You can see timing benchmark results of the three...

> thoughts on usefulness? I can't personally advocate for its usefulness, but I suppose it's needed if a user has weights _and_ wants to use a tuple `axis`. It sounds...

@mckib2 did we see this before?

Oops, @mckib2 has been working on one. Perhaps this can be added as another example.

Merging main + re-building docs so I can review.

@AtsushiSakai this is great! I made suggestions in https://github.com/AtsushiSakai/scipy/pull/3. Please consider merging that (with commit comment `[skip azp] [skip actions]`, perhaps), and then I think this is ready to merge!

Thanks @AtsushiSakai!