Fix wrong ZeroSumNormal logp expression

[lucianopaz]

Closes #6871

What is this PR about?

This PR fixes the ZeroSumNormal logp expression. It also changes the logp test to to compare the logp to the reference value for points drawn at random in the zero sum manifold and not only at the mode.

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 👇

Bugfixes

• Fix the logp expression for the ZeroSumNormal distribution

---

📚 Documentation preview 📚: https://pymc--6872.org.readthedocs.build/en/6872/

[codecov]

Codecov Report

Merging #6872 (e2ebb80) into main (9ee5d84) will increase coverage by 0.01%.
The diff coverage is 100.00%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #6872      +/-   ##
==========================================
+ Coverage   92.05%   92.07%   +0.01%     
==========================================
  Files          96       96              
  Lines       16445    16446       +1     
==========================================
+ Hits        15139    15143       +4     
+ Misses       1306     1303       -3     

Files Changed	Coverage Δ
pymc/distributions/multivariate.py	92.56% <100.00%> (+<0.01%)	⬆️

... and 1 file with indirect coverage changes

[lucianopaz]

I'll go ahead and merge