Speedup scalar Op python implementations

[ricardoV94]

Some cleanup of unused code in scalar/math.py.

Also speedup a bit the perform method by avoiding allocating intermediate variables and attribute acces for a helper function

---

📚 Documentation preview 📚: https://pytensor--1169.org.readthedocs.build/en/1169/

[codecov]

Codecov Report

Attention: Patch coverage is 53.33333% with 21 lines in your changes missing coverage. Please review.

Project coverage is 82.26%. Comparing base (4fa9bb8) to head (5d340e2).
Report is 184 commits behind head on main.

Files with missing lines	Patch %	Lines
pytensor/scalar/math.py	37.50%	20 Missing ⚠️
pytensor/scalar/basic.py	88.88%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1169      +/-   ##
==========================================
- Coverage   82.27%   82.26%   -0.01%     
==========================================
  Files         186      186              
  Lines       48066    47962     -104     
  Branches     8633     8630       -3     
==========================================
- Hits        39546    39456      -90     
+ Misses       6360     6347      -13     
+ Partials     2160     2159       -1     

Files with missing lines	Coverage Δ
pytensor/tensor/inplace.py	100.00% <ø> (ø)
pytensor/tensor/math.py	91.87% <100.00%> (+<0.01%)	⬆️
pytensor/tensor/xlogx.py	78.37% <100.00%> (-3.02%)	⬇️
pytensor/scalar/basic.py	80.59% <88.88%> (+0.04%)	⬆️
pytensor/scalar/math.py	87.76% <37.50%> (+0.65%)	⬆️

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Armavica]

Looks good, but there are two more st_impl in pytensor/tensor/xlogx.py: worth removing them too in this PR?

[ricardoV94]

definitely, missed those

[ricardoV94]

I removed the Chi2SF Scalar Op altogether. There is a tensor helper that still works with a future warning to just use gammaincc instead. This is not a standard scipy special function, and it doesn't make sense to support a redundant Op. It didn't even have gradients, where gammaincc does.