Global deformation for space mapping

[sblauth]

The newly introduced global deformation (see #222) should be used for space mapping for shape optimization (this makes the space mapping also work with remeshing).

[sblauth]

This cannot be done at the moment due to #320, where a bug with PETSc and the transfer matrix has been found.

[sblauth]

In principle, this can now be done. However, therre is another problem - if remeshing and the global deformation should be usable for the space mapping, then the coarse model problem (and parameter extraction) also require the same remeshing interface as the shape optimization problem. This then gives some further problems down the road, as at the moment, the coarse model is used to define the parameter extraction etc.

So this is a bigger change than first expected.