Add test-pypy3.8 test-pypy3.9 test-pypy3.10

[ecerulm]

Currently there are nox session for test-3.8, test-3.9, test-3.10, test-3.11 and test-pypy.

Then test-pypy is should in my opinion be split into test-pypy3.8, test-pypy3.9, and test-pypy3.10 since that is already done in CI. It easier to run multiple pypy locally in the laptop if those nox sessions are explicit.

With this PR is possible to run the nox sessions for all pypy versions like so

brew install pyenv 
brew install nox

pyenv install pypy3.8
pyenv install pypy3.9
pyenv install pypy3.10
pyenv install 3.8
pyenv install 3.9
pyenv install 3.10
pyenv install 3.11


#zsh
path=(
$(pyenv prefix 3.8)/bin
$(pyenv prefix 3.9)/bin
$(pyenv prefix 3.10)/bin
$(pyenv prefix 3.11)/bin
$(pyenv prefix pypy3.8)/bin
$(pyenv prefix pypy3.9)/bin
$(pyenv prefix pypy3.10)/bin
$path
)

nox -rs format lint mypy test-3.8 test-pypy3.8 test-3.9 test-pypy3.9 test-3.10 test-pypy3.10 

The rational behind this PR is to be able to run the pypy tests locally in my computer for pypy3.8, pypy3.9 and pypy3.10 without having to delete the .nox/pypy in between. Also I think for consistency with the cpython tests where there is a separate nox session for each one.

This PR has #3286 as prerequisite though.

[pquentin]

Thanks! I think lint is failing? You will need to run nox -rs format.

[pquentin]

Thanks! LGTM.

[pquentin]

This would be interesting to understand why we have so many issues with PyPy. Any ideas?

[ecerulm]

Well, I actually noticed that test-pypy fails very often under pypy3.9 on the CI, and I wanted to investigate more, but then I realized I could not test locally pypy3.8 and pypy3.9 hence this PR.

I don't really understand yet, why the test under pypy3.9 are sometimes very very slow (it takes more than 30 minutes to get to 50% of the test suite) , I want to check if that is something that happens only on the GItHub Actions / CI or if it is something that I can replicate with pypy3.9 locally.

I just wrote an issue to track the the pypy on ci problems: #3305

[ecerulm]

This would be interesting to understand why we have so many issues with PyPy. Any ideas?

@pquentin , I created PR #3308 which I think solves (or sidesteps) the issues with pypy3.9 in particular.

[sethmlarson]

@ecerulm Can you resolve the conflicts on this PR with the 3.9 one that I just merged?

[pquentin]

Comments on your changes to the contributing docs.