Use the same finite element for mesh deformation and stokes velocity

[gassmoeller]

The mesh deformation is (most often) caused by the Stokes velocity inside the domain, but we currently project the Stokes velocity into a Q1 space to compute mesh deformations. We should test using the same element as for the Stokes velocities. This will require changing the mapping that is currently used with mesh deformation from a linear MappingQ1Eulerian to a higher order MappingQEulerian.