Clean up scheme folder

[Øyvind Evju]

Remove schemes that don't work, combine schemes that are essentially the same, and use a naming convention.

[Øyvind Evju]

This would be easiest if we could improve regression tests, so we could keep a (smaller) batch of verified schemes.

[Martin Sandve Alnæs]

Issue #30 was marked as a duplicate of this issue.

[Martin Sandve Alnæs]

Here's a new variation: Keep supported schemes in cbcflow.schemes.*, and move experimental schemes to cbcflow.schemes.experimental.*. This makes experimental schemes available through a cbcflow import but keeps them out of the way and clearly marked. The experimental schemes should not be imported from cbcflow.*.

The tests already contain some lists of schemes, they can be updated to match just the supported schemes by default. Further improvement of the tests is another (set of) issue(s).

[Martin Sandve Alnæs]

Partially done: @oyvinev can you decide which schemes to keep in official? These are the ones that will be available through cbcflow.*, although I suggest we keep importing all schemes into cbcflow.schemes.*.

[Martin Sandve Alnæs]

I think we should stick with the same naming convention used in fields and demos now, i.e. cbcflow/schemes/official/IPCS.py contains class IPCS etc. If we're happy with the scheme class names, just rename the files to match. I don't like the mix of CamelCaps and underscore_naming though.