Tweak rewrites in infer_shape dataset

[ricardoV94]

This PR removes some inc_subtensor rewrites from infer_shape database. These don't simplify shape inference as the shape of an inc_subtensor is always the shape of the input, regardless of whether it is useless or not. They are also not very common to begin with?

Instead we add the useless_switch rewrite. We want a subset of rewrites that quickly simplify the graph and reason about the shapes introduced by infer_shape

---

📚 Documentation preview 📚: https://pytensor--1563.org.readthedocs.build/en/1563/

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.53%. Comparing base (b55c473) to head (666fe9c).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1563      +/-   ##
==========================================
- Coverage   81.53%   81.53%   -0.01%     
==========================================
  Files         230      230              
  Lines       53079    53078       -1     
  Branches     9425     9425              
==========================================
- Hits        43279    43278       -1     
  Misses       7365     7365              
  Partials     2435     2435              

Files with missing lines	Coverage Δ
pytensor/tensor/rewriting/basic.py	95.86% <100.00%> (+<0.01%)	⬆️
pytensor/tensor/rewriting/subtensor.py	90.25% <ø> (-0.03%)	⬇️

🚀 New features to boost your workflow:

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