Compute pushforward (Rop) via double application of pullback (Lop) and fix Scan and Max gradient bugs

[ricardoV94]

Extracted this from #788 as it's a pretty nice feature to have. This should allow Rop of basically any graph, whereas the specialized support of Rop was pretty patchy.

Also fixes bugs in Scan L_op and Max R_op revealed by the changes / new tests.

Related to #182
Closes #6

Not using the new grad names yet on user facing functions and docs, let's do that for the next major release

---

📚 Documentation preview 📚: https://pytensor--1207.org.readthedocs.build/en/1207/

[ricardoV94]

float32 tests fail, shocker...

[codecov]

Codecov Report

Attention: Patch coverage is 85.07463% with 10 lines in your changes missing coverage. Please review.

Project coverage is 82.28%. Comparing base (2aecb95) to head (4136161).
Report is 5 commits behind head on main.

Files with missing lines	Patch %	Lines
pytensor/tensor/math.py	46.15%	4 Missing and 3 partials ⚠️
pytensor/gradient.py	90.90%	2 Missing and 1 partial ⚠️

❌ Your patch status has failed because the patch coverage (85.07%) is below the target coverage (100.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1207      +/-   ##
==========================================
+ Coverage   82.25%   82.28%   +0.03%     
==========================================
  Files         186      186              
  Lines       47981    47987       +6     
  Branches     8630     8629       -1     
==========================================
+ Hits        39465    39486      +21     
+ Misses       6356     6347       -9     
+ Partials     2160     2154       -6     

Files with missing lines	Coverage Δ
pytensor/compile/builders.py	88.66% <100.00%> (ø)
pytensor/scan/op.py	84.62% <100.00%> (-0.08%)	⬇️
pytensor/gradient.py	78.55% <90.90%> (+0.92%)	⬆️
pytensor/tensor/math.py	92.00% <46.15%> (+0.12%)	⬆️

... and 2 files with indirect coverage changes

[aseyboldt]

Looks good :-)