Mixed velocity-vorticity transport

[tommbendall]

This PR introduces a new Augmentation object, which allows an equation set to be substituted or augmented for a particular stage. The example I have implemented is the mixed velocity-vorticity transport of this paper

The cool thing about this transport is that it is more accurate than other velocity transport schemes that we use, but @gwimmer also came up with a way of incorporating an SUPG scheme into it. I have been testing this in shallow-water models which take large time steps (such that 0.5 < c < 1.0) and it gives good results. Unfortunately I haven't found anything completely stable for c > 1.0 yet ...

Still to do:

• unify and simplify approach to subcycling (maybe in a different PR)
• provide a transport test using the augmentation
• provide an integration test using my new favourite settings
• decide on optimal solver parameters
• test 3D scheme
• docstrings and tidying