remove (or rename) type alias volume_dim in evolution/elliptic system structs

[kidder]

The dimension of the computational domain on which a system is solved is independent of that system. (For example I can solve 3-d MHD equations on a 1-d grid.) So the type alias should be removed or renamed if it is useful to keep a dimension for the system (i.e. the spatial dimension of the tensors in the evolution/elliptic equations defining the system)

[nilsvu]

So should we keep the system::volume_dim and add Metavariables::domain_dim? That‘s what your description above sounds like - the metavars keep track of the computational domain dimension and the system of the physical volume dimension. domain_dim is already used by the Interpolation code. Not sure if we need the system::volume_dim at all though.

[nilsdeppe]

For subcell I'm doing metavariables::mesh_dim for the dimensionality of the mesh and metavariables::volume_dim for the tensors/system.