Remove custom standard RV Op(s) and add remaining "standard" NumPy RVs

[Smit-create]

Fixes #1358

Here are a few important guidelines and requirements to check before your PR can be merged:

• There is an informative high-level description of the changes.
• The description and/or commit message(s) references the relevant GitHub issue(s).
• pre-commit is installed and set up.
• The commit messages follow these guidelines.
• The commits correspond to relevant logical changes, and there are no commits that fix changes introduced by other commits in the same branch/BR.
• There are tests covering the changes introduced in the PR.

[brandonwillard]

We really need to do something about the existing StandardNormalRV (i.e. #530), because I believe the new class approach was taken due to its convenience with regard to RandomStream (i.e. it works with the current RandomStream.__getattr__ implementation). See #528.

In general, we don't want to add new Op classes if we don't have to, especially in cases like this where the new Op type implies that its corresponding Apply nodes will have specific inputs, but there is nothing enforcing this constraint. One reason this approach is problematic: an Apply using a StandardNormalRV Op can be rewritten to have non-zero mean and/or non-unit variance. This implies that StandardNormalRV is redundant at best, but can also be misleading. (Likewise, adding input constraints to fix this wouldn't help our graph representations in any foreseeable way; if anything, it would unnecessarily complicate things.)

That said, we should first see what can be done about making RandomStream work with simple definitions like standard_normal = normal(0, 1) and the partial functions.

[Smit-create]

Ah, the failure on CI is hard to reproduce on my machine. It says:

FAILED tests/link/jax/test_random.py

But pytest does pass for that on my setup.

[brandonwillard]

But pytest does pass for that on my setup.

This is often due to version mismatches between local environments and CI.

I think this issue appeared in another PR as well, so you can add a commit marking it as xfail for the time being and we'll get back to it later/elsewhere.

[brandonwillard]

The issue should be fixed in #1464.

[codecov]

Codecov Report

Merging #1461 (60445b7) into main (c433071) will increase coverage by 0.01%.
The diff coverage is 91.42%.

❗ Current head 60445b7 differs from pull request most recent head ffa4d81. Consider uploading reports for the commit ffa4d81 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1461      +/-   ##
==========================================
+ Coverage   74.86%   74.87%   +0.01%     
==========================================
  Files         194      194              
  Lines       50107    50129      +22     
  Branches    12098    12097       -1     
==========================================
+ Hits        37514    37536      +22     
  Misses      10266    10266              
  Partials     2327     2327              

Impacted Files	Coverage Δ
aesara/compile/profiling.py	74.55% <ø> (ø)
aesara/link/jax/dispatch/random.py	100.00% <ø> (ø)
aesara/tensor/basic.py	89.99% <ø> (ø)
aesara/tensor/elemwise.py	88.07% <ø> (ø)
aesara/tensor/var.py	87.82% <50.00%> (ø)
aesara/tensor/rewriting/math.py	86.11% <60.00%> (ø)
aesara/link/jax/dispatch/scalar.py	96.72% <100.00%> (ø)
aesara/link/numba/dispatch/elemwise.py	97.12% <100.00%> (ø)
aesara/scalar/basic.py	79.17% <100.00%> (-0.01%)	⬇️
aesara/tensor/inplace.py	100.00% <100.00%> (ø)
... and 2 more

[brandonwillard]

I added a fix for the use of partial functions in RandomStream and rebased. This should be good to merge once the tests pass.