Add a finite volume method for ScalarEq

Proof of principle.

Introduced a (suboptimal) extra FVElement type and added another type parameter to Mesh. Added a ref test for the advection eq with periodic sine wave.

This PR also did some refactoring of the ScalarEq project:

• remove all occurrences of the old broadcast methods (now using broadcast_*_2! or newer)
• introduced some abstractions for Mesh to compactify time step computation

So far only 1d is implemented!