Misc. aesara.tensor.random updates

[brandonwillard]

This PR provides multiple aesara.tensor.random updates (e.g. removes the unnecessary Observed Op, exposes RandomVariables via aesara.tensor.random, etc.). It also allows the use of numpy.dtype instances (e.g. numpy.int64) as dtype arguments during tensor and scalar construction, as well as aesara.tensor.cast.

[ricardoV94]

LGTM.

On a separate note, none of these random methods show up in the docs do they?

[brandonwillard]

On a separate note, none of these random methods show up in the docs do they?

I don't recall updating the list manually, but definitely seems like something that should be generated automatically.

[codecov]

Codecov Report

Merging #436 (ccede9b) into master (1c11dd4) will decrease coverage by 0.00%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #436      +/-   ##
==========================================
- Coverage   72.45%   72.44%   -0.01%     
==========================================
  Files         173      173              
  Lines       55651    55640      -11     
==========================================
- Hits        40320    40309      -11     
  Misses      15331    15331              

Impacted Files	Coverage Δ
aesara/link/numba/dispatch.py	98.09% <ø> (-0.01%)	⬇️
aesara/scalar/basic.py	84.68% <100.00%> (+<0.01%)	⬆️
aesara/tensor/basic.py	88.73% <100.00%> (+<0.01%)	⬆️
aesara/tensor/random/basic.py	100.00% <100.00%> (ø)
aesara/tensor/random/op.py	99.32% <100.00%> (-0.06%)	⬇️
aesara/tensor/type.py	93.35% <100.00%> (ø)