Prevent unbound trace due to type hints

[thomasaarholt]

There are currently two type errors in the Model class, preventing smooth usage with type checkers (like in VSCode):
This one became present in recent versions of pyright:

# script.py
import pymc as pm
import arviz as az

with pm.Model() as m: # <-- (...) does not implement __enter__ 
    pm.Normal("a", mu=0, sigma=1)
    trace = pm.sample()

az.plot_posterior(trace)

and before that one, this one has been there for a long time:

# script.py
import pymc as pm
import arviz as az

with pm.Model() as m:
    pm.Normal("a", mu=0, sigma=1)
    trace = pm.sample()

az.plot_posterior(trace) # <-- 9:19 - error: "trace" is possibly unbound (reportUnboundVariable)

This PR implements two small changes:

1. Sets the type of the instance variable in the ContextMeta __call__ method to "Model". This fixes the first error.
2. Sets the return type of the context manager used by with pm.Model() to return None, which means that it won't be catching errors.

In the first fix, instance should be a resultant type that the ContextMeta is a metaclass for. I'm not sure how if a type hint that says "returns the child of a metaclass" exists. But I think this is safe, since ContextMeta isn't used as a metaclass anywhere else, and I'm not going for perfect type coverage, just to make it easy for people to code with autocompletion.

Checklist

• Explain important implementation details 👆
• Make sure that the pre-commit linting/style checks pass.
• Link relevant issues (preferably in nice commit messages)
• Are the changes covered by tests and docstrings?
• Fill out the short summary sections 👇

Bugfixes

• Fixed pm.Model not being interpreted as a context manager by pyright
• Fixed unbound type errors due to objects being created inside the pm.Model context

---

📚 Documentation preview 📚: https://pymc--6809.org.readthedocs.build/en/6809/

[codecov]

Codecov Report

Merging #6809 (ce3831e) into main (f67ff8b) will increase coverage by 0.00%.
The diff coverage is 50.00%.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #6809   +/-   ##
=======================================
  Coverage   92.02%   92.02%           
=======================================
  Files          95       95           
  Lines       16261    16262    +1     
=======================================
+ Hits        14964    14965    +1     
  Misses       1297     1297           

Impacted Files	Coverage Δ
pymc/model.py	90.95% <50.00%> (+0.01%)	⬆️

[thomasaarholt]

Tagging @michaelosthege on this, since he reviewed my other typing PR. I've also edited the original post in this PR and deleted my now superfluous comments.

I realise it's summer holidays for many, and I'm in no rush to get it merged, but don't want this to get lost.

[ricardoV94]

Thanks @thomasaarholt!