Usability and improve quality of core code

[rec]

Why?

We have a large amount of fairly fresh code, much of which is not covered by any unit test and is only partly documented.

Before launch, we need to review the core code for problems, document, add types, and increase test coverage.

How?

Each Python file eventually needs to:

• Be reviewed for errors and issues
• Be documented (doc comments for functions, classes, members, methods, and parameters)
• Have "good" typing (perfect typing is impractical)
• Have "good" test coverage (see above)

What

The plan is to start with superduperdb/core and continue down from there.

The rough order is:

• serialization
• artifacts
• containers
• documents
• ?

Sub-issues

• Audit all "type-ignores", fix errors.
• Core core: simplify and document configuration system #773
• Core core: remove dead code #774
• Core code: correct and complete all public doc-strings #780
• Core code: document binder playground to make it clear what is going on #801
• Core code: mongomock, make it easy for users to get started #961
• Core code: Improve and test device management #945

[nenb]

I'm going to suggest also tracking mypy-related issues here as well. If this is not appropriate please move elsewhere.

Current mypy issues:

• Core code: Audit and either fix, or explain, all typing errors #738
• Core code: Much of mypy has been disabled #524
• Core code: Increase the strictness of mypy #807

[blythed]

New issues:

• configure which collections are in use, and move cdc configuration back to the top level (no CDC_COLLECTION_LOCKS).

[blythed]

Addressed by #1243