Add missing type hint imports, as suggested by mypy.

[obi1kenobi]

A few minor fixes for type hint issues reported by mypy:

arviz/stats/stats.py:1133: error: Name 'Dict' is not defined
arviz/stats/stats.py:1133: note: Did you forget to import it from "typing"? (Suggestion: "from typing import Dict")
arviz/data/io_pyro.py:21: error: Name 'Optional' is not defined
arviz/data/io_pyro.py:21: note: Did you forget to import it from "typing"? (Suggestion: "from typing import Optional")
arviz/data/io_numpyro.py:19: error: Name 'Optional' is not defined
arviz/data/io_numpyro.py:19: note: Did you forget to import it from "typing"? (Suggestion: "from typing import Optional")

I also replaced use of builtins.callable with typing.Callable in type hints since mypy doesn't consider the lowercase version a valid type:

arviz/data/io_pyro.py:22: error: Function "builtins.callable" is not valid as a type
arviz/data/io_pyro.py:22: note: Perhaps you need "Callable[...]" or a callback protocol?

Unfortunately, this isn't a comprehensive set of type hint fixes -- mypy is still quite unhappy even with these changes applied. I just figured something is better than nothing 😃

[codecov]

Codecov Report

Merging #1397 into master will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #1397   +/-   ##
=======================================
  Coverage   91.63%   91.63%           
=======================================
  Files         105      105           
  Lines       10965    10967    +2     
=======================================
+ Hits        10048    10050    +2     
  Misses        917      917           

Impacted Files	Coverage Δ
arviz/data/io_numpyro.py	95.27% <100.00%> (+0.03%)	⬆️
arviz/data/io_pyro.py	97.35% <100.00%> (+0.01%)	⬆️
arviz/stats/stats.py	96.31% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 24ce65a...569d052. Read the comment docs.

[ahartikainen]

We currently don't implement or require type hints, but I think it could be a good addition to our lib. Is there yet a good solution for numpy / xarray related problems (e.g. we would not need to use Any)

[obi1kenobi]

Is there yet a good solution for numpy / xarray related problems (e.g. we would not need to use Any)

It seems like some solid progress is being made within the packages themselves: numpy/numpy#7370 It doesn't seem like it's super thorough yet, but it seems quite actively being worked on so it'll hopefully get even better quickly.