Ralf W. Grosse-Kunstleve

This looks useful to me, the change is relatively small, and the added documentation is well done. Caveat: I don't know much about sphinx and stubgen (don't use it in...

Thanks for updating this PR! It turns out, this PR breaks one test (out of millions). The breaking test involves this unexpected diff: ``` @@ -6,4 +6,8 @@ def __init__(self,...

> it will require adding another option and I don't love that this pull request would then increase the number of options from two to four. I feel the same,...

I'll close and reopen after a few seconds to trigger a CI rerun, to see if that shakes out the pip errors (I saw them in another PR yesterday but...

I glanced back and forth a few times and it looks great to me. (Unfortunately the force pushes make it difficult to re-review. For the future, please just add commits,...

No it's fine, no need to spend extra time. I just submitted the PR for the global testing. If there are no hiccups (rare), results tomorrow morning (Pacific).

Global test results look good! (Sorry I had to spend more time than usual deflaking). I'm thinking, for completeness it would be ideal to add a 4th new test with...

The force push was for `git rebase master`.

Tagging @jbms: is there a chance that you could look at this PR (and my uncertainties under https://github.com/pybind/pybind11/pull/3798#discussion_r829737887)?

General comment: we (mostly CPython) is sending people into traps by giving the impression that restarting the interpreter is safe, well-defined behavior. It simply isn't. It probably never will be....