It is handled by a separate class now

[[Imported from SVN: r10118]]

[[Imported from SVN: r10117]]

[[Imported from SVN: r10116]]

[[Imported from SVN: r10115]]

[[Imported from SVN: r10114]]

[[Imported from SVN: r10112]]

[[Imported from SVN: r10106]]

Because I am using the Debian version of IPOpt now.  Finally it works!
But the Debian version can only use "mumps", not the default "ma27".

[[Imported from SVN: r10105]]

[[Imported from SVN: r10104]]

[[Imported from SVN: r10103]]

[[Imported from SVN: r10101]]

[[Imported from SVN: r10100]]

[[Imported from SVN: r10099]]

[[Imported from SVN: r10097]]

[[Imported from SVN: r10092]]

[[Imported from SVN: r10091]]

Getting intersections from the element has been deprecated for a long time,
because the precise intersection one gets does not only depend on the element,
but also on the grid view.

[[Imported from SVN: r10090]]

[[Imported from SVN: r10089]]

[[Imported from SVN: r10088]]

[[Imported from SVN: r10087]]

Conceptually cleaner, and makes for simpler code.  Most importantly, we will want
to pass the Basis::LocalView to the local assemblers.  For that we need the actual
basis, not just the grid view and local finite element.

[[Imported from SVN: r10082]]

At least the harmonicmaps.cc example still works as before.
[caveat: only tested without HAVE_MPI]
Internally, at various places in the code, a dune-fufem basis is still constructed
from the dune-functions one.  These parts can be ported one by one, as dune-fufem
slowly replaces its own bases by the ones from dune-functions.

[[Imported from SVN: r10080]]

[[Imported from SVN: r10077]]

[[Imported from SVN: r10076]]

[[Imported from SVN: r10070]]

[[Imported from SVN: r10069]]

[[Imported from SVN: r10068]]

Which is the type returned by the derivativeOfProjection method.

[[Imported from SVN: r10067]]

Patch by Jonathan Youett

[[Imported from SVN: r10056]]

Patch by Jonathan Youett

[[Imported from SVN: r10055]]

This is a fall-out from the recent move to use all the parallel code
only when HAVE_MPI is set.  Very ugly, but I still don't know any other
way to have P3 spaces.

Patch by Jonathan Youett

[[Imported from SVN: r10054]]

Patch by Jonathan Youett

[[Imported from SVN: r10053]]

Use std::vector instead of C-array as clang does not support initialization of a C-array by a variable.

Patch by Jonathan Youett

[[Imported from SVN: r10052]]

Patch by Jonathan Youett

[[Imported from SVN: r10051]]

I don't see an elegant way to make this work for a P3 space, even on a single
processor.  Since I need P3 measurements real soon now, I just disable the
entire code.  Not nice.

[[Imported from SVN: r10047]]

The multigrid hierarchy is built in two different ways, depending on whether
we use a P1NodalBasis for the discretization, or anything else.  Previously,
this switch was made using the preprocessor.  The patches changes the code
to use std::is_same instead.

[[Imported from SVN: r10046]]

This is much cleaner than handing over the Grid only, and then trying
to guess the basis using the preprocessor.  After all, the
RiemannianTrustRegionSolver calls the assembler, so it may know the
function space basis.

[[Imported from SVN: r10035]]

By introducing a factory class that produces GlobalMapper and LocalMapper
types for a given dune-fufem-style function space basis.

[[Imported from SVN: r10034]]

[[Imported from SVN: r10033]]

[[Imported from SVN: r10032]]