ClementInterpolation enhancements

[MiroK]

Here are a few ways to improve the ClementInterpolation functionality:

• Add an additional parameter, e.g. user's dx, that would allow to exclude certain cells from construction of the interpolant.
• ClementInterpolation with $CG$ spaces could work also for quadratic order (across all dimensions). With polynomial order > 2 the interior dofs of entities (cell, facets, ...) would be set to same value and are therefore of little use. $CG_2$ could be added. It would require a different averaging operator.
• Consider extensions to $BDM$, $RT$, $NED$ elements. Might be useful for evaluating normal/tangential fluxes