Allow logcdf and icdf inference

[ricardoV94]

This PR allows inference of logcdf and icdf variables (similar to how logp was already possible)

import pymc as pm

dist = pm.Normal.dist().exp()

print(pm.logcdf(dist, 1.0).eval())  # -0.6931471805599453
print(pm.icdf(dist, 0.7).eval())  # 1.6894457434840042

Important

This PR first 3 commits now belong to #6599, only the last one should be reviewed here

[codecov]

Codecov Report

Merging #6597 (360ef6e) into main (2fcce43) will decrease coverage by 18.59%.
The diff coverage is 58.24%.

❗ Current head 360ef6e differs from pull request most recent head 0636f4e. Consider uploading reports for the commit 0636f4e to get more accurate results

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #6597       +/-   ##
===========================================
- Coverage   92.03%   73.44%   -18.59%     
===========================================
  Files          93       92        -1     
  Lines       15746    15591      -155     
===========================================
- Hits        14492    11451     -3041     
- Misses       1254     4140     +2886     

Impacted Files	Coverage Δ
pymc/logprob/transforms.py	57.14% <22.22%> (-39.00%)	⬇️
pymc/distributions/mixture.py	59.19% <33.33%> (-36.24%)	⬇️
pymc/logprob/mixture.py	31.28% <33.33%> (-67.24%)	⬇️
pymc/logprob/scan.py	19.79% <33.33%> (-77.42%)	⬇️
pymc/logprob/tensor.py	41.52% <33.33%> (-39.84%)	⬇️
pymc/logprob/basic.py	80.00% <50.00%> (ø)
pymc/logprob/cumsum.py	42.42% <50.00%> (-57.58%)	⬇️
pymc/logprob/abstract.py	89.28% <78.94%> (-8.31%)	⬇️
pymc/distributions/bound.py	46.15% <100.00%> (-53.85%)	⬇️
pymc/distributions/continuous.py	97.69% <100.00%> (-0.01%)	⬇️
... and 7 more

... and 36 files with indirect coverage changes

[ricardoV94]

Marking as a draft as this PR now includes the commits of #6599 which should be merged first. Only the last one matters here (and can be reviewed alone)

[ricardoV94]

Ready for review