This is a preparation to allow to select the TargetSpace at run-time.

Needed for gradient-flow applications.

You never know, we may be doing dim!=dimworld grids eventually.

Of course we always need unitvector.hh.  vtkadapter.hh, on the other hand, only
exists in dune-functions 2.4.

According to Hanne, this is needed to turn 1d harmonic map heat flow into true
curve-shortening flow.

Hardwiring the weighting here is very hacky.  Something more flexible and
general will need to be implemented eventually.

Apparently, nobody had used the evaluateDerivativeLocal method before.

This is needed even though we don't actually use MPI here.  Dune aborts with a
run-time error otherwise.

This are needed by the gradientflow code to implement the implicit Euler method
for an L2 gradient flow.

This is an initial version; it currently only works for a very limited set of
cases.  I tested it for triangle grids, and that works.

It is so uncompileable that I am not able to fix it quickly.  Since I don't
actually need it right now I let it wait for another day.

And make that code path work again!

This is needed for the new gradientflow application.  The standard 'distance' method
doesn't cut it, because it is not differentiable near zero.  Therefore, even the
differentiable expression 'distance*distance' will fail to be differentiable for
an automatic-differentiation system.  I think in the long run we should replace
'distance' by 'distanceSquared' everywhere it is used.

Unfortunately, this patch is hacky: it only adds the method for the double/adouble
combination.

That should work when calling the executable from 'build-cmake/src'.

This initial commit simply brings a cleaned-up version of the harmonicmaps program.
The first step will be to make it compute curve-shortening flow on the sphere.

Previously, the NonplanarCosseratShellEnergy had the normal field of the unit sphere
hard-wired in the code.  With this patch, using general triangulated surface as
reference geometries is possible.

For some reason clang doesn't like it, and we are not using OpenMP anyway.