Revert numpy warnings workaround

[maresb]

This workaround was originally written for Aesara in aesara-devs/aesara#980 to suppress spurious warnings when Aesara loads Numpy under Windows. It was subsequently ported to run here on pymc in #5956. It has been subsequently fixed upstream in Aesara in a superior way by aesara-devs/aesara#1050. Therefore, this workaround is obsolete, so I'm reverting it.

What is this PR about?
...

Checklist

• Explain important implementation details 👆
• Make sure that the pre-commit linting/style checks pass.
• Link relevant issues (preferably in nice commit messages)
• Are the changes covered by tests and docstrings?
• Fill out the short summary sections 👇

Major / Breaking Changes

• None (hopefully!)

Bugfixes / New features

• None

Docs / Maintenance

• Cleans up unnecessary code 🚀

Tests:

AFAIK there's no test here for warnings. But there is on Conda-Forge, so any problems on a release would be caught very quickly.

[codecov]

Codecov Report

Merging #6025 (08b1079) into main (9f93b3e) will increase coverage by 3.00%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##             main    #6025      +/-   ##
==========================================
+ Coverage   86.28%   89.28%   +3.00%     
==========================================
  Files          72       72              
  Lines       12905    12884      -21     
==========================================
+ Hits        11135    11504     +369     
+ Misses       1770     1380     -390     

Impacted Files	Coverage Δ
pymc/__init__.py	100.00% <ø> (+31.57%)	⬆️
pymc/step_methods/hmc/base_hmc.py	89.76% <0.00%> (-0.79%)	⬇️
pymc/model.py	88.20% <0.00%> (+0.13%)	⬆️
pymc/backends/base.py	86.44% <0.00%> (+0.73%)	⬆️
pymc/step_methods/slicer.py	96.20% <0.00%> (+1.26%)	⬆️
pymc/backends/report.py	91.04% <0.00%> (+4.47%)	⬆️
pymc/step_methods/arraystep.py	94.32% <0.00%> (+7.80%)	⬆️
pymc/data.py	80.08% <0.00%> (+12.19%)	⬆️
pymc/sampling.py	82.48% <0.00%> (+12.33%)	⬆️
pymc/step_methods/metropolis.py	83.36% <0.00%> (+43.65%)	⬆️

[maresb]

Ping @michaelosthege since he coauthored the commit I'm reverting.

[michaelosthege]

I assume you confirmed with a recent version that the import is now indeed silent?

My environment are not on the latest version yet..

[maresb]

Not really, that would require a Windows machine.

It's tested fairly thoroughly at the level of the Aesara feedstock on Conda-Forge. Although there is a very slight possibility that PyMC does some funny import order and breaks the Aesara fix somehow.

If you want, I think I could rig up a draft PR on the PyMC feedstock and redirect it to this branch in order to verify it on one of the Conda-Forge Windows VMs. (I can't do that right away though.)