ENH: Flatmaps for cortex

[larsoner]

See the WIP FreeSurfer PR where there is an implementation we can work with:

nipy/PySurfer#269

The tricky part is that the vertices of the flatmap do not directly map onto the vertices of the cortex. I'm not sure why it is this way. I wonder for us if a better implementation would actually be to resample the flatmaps themselves (in the spherical space maybe?) to make them have the same effective vertices or something.

cc @johnsam7 this is probably a prerequisite for being able to do it for cerebellum

I'll probably take a look to see how difficult this would be