Remove initval from dist() API and add docstrings, type hints, tests

[michaelosthege]

This PR makes initial_values a property with docstring and type hints. It also adds some more type hints and docstrings to related functions and properties.

For the UNSET constant I added its own type, because inspecting signatures with UNSET kwargs (for example pm.Normal) results in confusing output, as demonstrated by this example:

>>> UNSET = object()
>>>
>>> def say_hello(name=UNSET):
...     print("Hi!" if name is UNSET else f"Hello {name}.")
...
>>>
>>> help(say_hello)
Help on function say_hello in module __main__:

say_hello(name=<object object at 0x0000018BD4226150>)
#                👆

With the new repr it just becomes

Help on function say_hello in module __main__:

say_hello(name=UNSET)

Depending on what your PR does, here are a few things you might want to address in the description:

• what are the (breaking) changes that this PR makes? None
• important background, or details about the implementation
• are the changes—especially new features—covered by tests and docstrings?
• linting/style checks have been run
• consider adding/updating relevant example notebooks
• right before it's ready to merge, mention the PR in the RELEASE-NOTES.md not relevant enough

[codecov]

Codecov Report

Merging #4913 (03b22c3) into main (06d1412) will increase coverage by 0.03%.
The diff coverage is 96.96%.

@@            Coverage Diff             @@
##             main    #4913      +/-   ##
==========================================
+ Coverage   73.12%   73.15%   +0.03%     
==========================================
  Files          86       86              
  Lines       13856    13868      +12     
==========================================
+ Hits        10132    10145      +13     
+ Misses       3724     3723       -1     

Impacted Files	Coverage Δ
pymc3/model.py	83.59% <94.11%> (+0.11%)	⬆️
pymc3/distributions/continuous.py	96.24% <100.00%> (-0.02%)	⬇️
pymc3/distributions/distribution.py	81.88% <100.00%> (+1.44%)	⬆️
pymc3/util.py	74.35% <100.00%> (+0.84%)	⬆️
pymc3/backends/report.py	89.51% <0.00%> (-2.10%)	⬇️
pymc3/parallel_sampling.py	88.21% <0.00%> (+1.01%)	⬆️