Improving Type Hinting

[john-science]

Description

This PR adds type hinting to an arbitrary subset of our code base; basically low-hanging fruit. This action was undertaken as part of this discussion: #493

The process taken to generate these type hints was relatively easy.

1. Install monkeytype.
2. Execute monkeytype run armi/path/to/test_whatever.py to determine type hints for functions.
   • This can be done on multiple test files, and all the type hints will be collated.
3. Execute monkeytype stub armi.path.to.whatever to print-to-screen type hints.

This process is easy to do. However, it is not perfect:

• Most of our unit test files cannot be run as-is using monkeytype, because they do no explicitly declare an App(), so without pytest they just fail.
• Many of the type hints monkeytype output that are our custom classes are only pseudo-correct. For instance, in many places in our codebase a method might return any subclass of one of our custom classes, and monkeytype doesn't handle OOP that well. (Conversely, it might be that Python's type hinting is not well designed to handle this very common use case).
• Some of the unit tests we run don't fully flex a given function, so 'monkeytype just doesn't have the info necessary to get the type hinting 100% right. I have found many partial errors.

It seems to me that monkeytype is a useful tool. So I am opening this PR just as an example of the benefits it can generate. But due to the issues I listed above, it was not an automatic enough process for me to just sweep the whole code base and commit everything.

[jakehader]

Hi @john-science, this looks very reasonable and thanks for the description of the steps you took to get monkeytype to work. I just have one question.

[jakehader]

Hi @john-science, this looks very reasonable and thanks for the description of the steps you took to get monkeytype to work. I just have one question.

How are you doing? Hilarious that I said I had a question and never asked it!