Previously, the number of multigrid transfer operators was always equal to
the number of grid levels minus one.  That worked out correctly when doing
first-order FE spaces.  However, for higher-order spaces, the up-most transfer
operator transfers from the higher-order space to a first-order space
_on the same grid level_.  But since the total number of transfer operators
was grid levels minus one, this implied that the coarsest grid leve 
was never used, and the coarse grid solver would operate on the second-coarsest
grid.

This patch adds the extra transfer operator in this case.  That means we now
call the coarse grid solver on smaller problems, which makes a noticeable
run-time difference.

[[Imported from SVN: r9964]]

[[Imported from SVN: r9963]]

Normally, the two different level grid views.  Previously we used the same one
to compute the localToGlobal renumbering for boths rows and columns.  Of course
that screws up the numbering.

I am not quite sure when that got introduced.  I may have gotten in fairly recently,
after the upstreaming of the GlobalIndexSet classs.

[[Imported from SVN: r9962]]

[[Imported from SVN: r9961]]

[[Imported from SVN: r9960]]

[[Imported from SVN: r9959]]

May happen when the matrix is singular.  UMFPack doesn't like that.

[[Imported from SVN: r9958]]

It tentatively uses the UMFPack direct solver, disregarding the obstacle.
If that leads to an energy-decreasing, admissible new iterate, then good.
Otherwise, it falls back to IPOpt.

[[Imported from SVN: r9957]]

[[Imported from SVN: r9956]]

[[Imported from SVN: r9955]]

[[Imported from SVN: r9954]]

Because it is not an index, it really is a mapper.

[[Imported from SVN: r9953]]

[[Imported from SVN: r9952]]

[[Imported from SVN: r9951]]

This is a first step towards having trust-regions with different scaling
in the different components.  Kinda paradox to introduce an explicit radius
in order to move away from a single radius.  But I want to pretend to
the outside (for the time being) that there is a single radius...

[[Imported from SVN: r9950]]

[[Imported from SVN: r9949]]

We may want to override options from the parameter file.

[[Imported from SVN: r9946]]

[[Imported from SVN: r9945]]

[[Imported from SVN: r9944]]

[[Imported from SVN: r9943]]

To really make sure I am doing the infamous L-shape example
like the people before me did I need point loads.  Being a
mathematician I am no particular fan of point loads, so they
may go out again eventually.

[[Imported from SVN: r9942]]

[[Imported from SVN: r9941]]

This is the interpolation method used by Ingo Münch and Wolfgang Müller for
Cosserat materials.  My implementation is only able to do all that I needed
for the illustrations in the GAMM Rundbrief article.  I'm sure there'll be
more fixes needed if you actually want to do finite elements with this.

[[Imported from SVN: r9940]]

For example, I'll use this for finite-strain elasticity problems.

Some of this should be in dune-fufem and/or dune-pdelab, but for the time
being it's easier to have it here.

[[Imported from SVN: r9939]]

[[Imported from SVN: r9938]]

This form allows to set the shear correction factor kappa.  It is still
not ruled out that we need to twiddle with this factor in order to
weaken the material a little bit.

[[Imported from SVN: r9937]]

[[Imported from SVN: r9936]]

[[Imported from SVN: r9935]]

[[Imported from SVN: r9934]]

[[Imported from SVN: r9932]]

[[Imported from SVN: r9931]]

[[Imported from SVN: r9930]]

Gauss-Seidel is simply to slow, the Newton convergence suffers too much.

[[Imported from SVN: r9929]]

[[Imported from SVN: r9928]]

[[Imported from SVN: r9927]]

That code was buggy, and we only didn't see that because the wrong sizes
were always overwritten by the correct ones before being used.
(That's so as long as you call a reducing method before 'scatter').

Removing the code leaves the situation as bad as before, but it
allows us to get rid of the nOwnedLocalEntity() method from the global
mapper.

[[Imported from SVN: r9926]]

In particular this means having to carry mappers for the local indices around.
This makes the code more verbose, but it also makes it easier to see what is
happening.  While coding this stuff I keep having doubts about whether all
the renumbering works correctly...

[[Imported from SVN: r9925]]

The global indexing infrastructure does not currently support P3 spaces.

[[Imported from SVN: r9924]]

[[Imported from SVN: r9923]]

The local copy was to get the thing developed and working.  Now that it is
working we upstream it into dune-grid.

[[Imported from SVN: r9922]]