Simplify

[di]

This PR ditches the ClassifierNode/ClassifierTree entirely (along with the code generation) since we don't actually need the tree hierarchy for PyPI.

Tests now run against the same classifiers/deprecated_classifiers objects that end up in the package.

Also, switches to CalVer.

[ewdurbin]

This is rad! I'm a big fan of the simplification, but have a couple nits:

• Can the ordering of the items trove_classifiers/__init__.py be enforced via test? I don't think it is crucial... would just help keep things sane. I'd say this is priority 1/10
• The structure of deprecated_classifiers is pretty vague. Now that it's not machine generated a comment describing it would probably get good mileage. priority 5/10, worth changing but not a blocker by any means.

[di]

Can the ordering of the items trove_classifiers/__init__.py be enforced via test?

If you have any idea how to do this, I'm all ears. But since these are sets/dicts respectively, it'll be hard to do w/o doing some static analysis of the file.

Maybe there's a flake8 plugin or something that enforces order for elements in unordered data structures?

[pradyunsg]

I mean... we can add a pytest test that reads the file, and checks if the set's contents are indeed correctly. It's OK to be slightly fragile at times.

PS: I'm only slightly crazy. :)

[di]

Created a new issue to capture this: #7.