Calculation of trigamma and polygamma functions for complex inputs

[LorenzoPeri17]

Closes issue #120 and #7410 in scipy.

I have started a PR to get the ball rolling. I have implemented the tests myself calculating raw input-output pairs via mpmath because I could not quite wrap my head the table generation system of xsref. Happy to help in this effort with some guidance :)

I would also be happy to help with the plumbing to get trigamma and poligamma plumbed into scipy.special if there is a need!

I have taken the liberty of making a bespoke trigamma function because it is the one that comes up regularly in my work (and it's the original reason why I went down the polygamma rabbit hole), so I could use the special handling of the roots ;)

[steppi]

Thanks @LorenzoPeri17! Don't worry about xsref, that's not the process contributors have been using to add tests. You can instead add tests to the xsf_tests folder https://github.com/scipy/xsf/tree/main/tests/xsf_tests. scipy_special_tests was created through a process of automatically generating tests for the existing scipy.special test cases from the SciPy repo, and all the infra around it really only makes sense for large scale automated test generation.

[fbourgey]

Thanks @LorenzoPeri17

CI is complaining because your code is not formatted.

You can run clang-format -i my_file.cpp to format using clang-format

[LorenzoPeri17]

Thanks for the comments! I just realized that I ran clang-format on the headers only, not the tests. Will run it again!

And thanks for the clarification @steppi about xsref! Is it then fine if I move the tests as the are (with the precomputed values from mpmath as tables) and move them to the xsf_tests folder?