We simply return a reference to the internal memory instead of copying the object.
This should save some copying when using FieldVector/FieldMatrix with non-trivial
types.  Also, it fixes crashes I have experienced with the ADOL-C automatic
differentiation toolbox [0].

I hope that this patch doesn't have any nasty side effects.  I didn't see any
so far.  Comments for this patch had been requested in FS 1449, and there have
been no objections.

[0] I have contacted the ADOL-C maintainer about this and he said the crashes
are not really bugs in ADOL-C, but that it is simply very difficult to write
a good operator-overloading AD program that can handle operator K returning
values.

The classes AlwaysFalse, AlwaysTrue reside in the deprecated header
dune/common/static_assert, yet they are frequently used to document
template code (see FS#1435 for details). As the C++11 standard library
does not provide a straight forward replacement for theses classes, we
offer C++11 style reimplementations. The replacements to_false_type,
to_true_type reside in the namespace Std and are thus easily
identifiable as Dune specifis.

Unfortunately, git-log doesn't use this file by default.  Only git-blame
and git-shortlog do.

Thanks to Elias Pipping for the hint about mailmap.

Add template parameter for the allocator of the internal std::vector.
All constructors are adadpted to forward a user chosen allocator
to the internal data stroage (std::vector).

Elias pointed out that one of the constructors got lost when merging
with the recent HEAD. This commit adds the missing constructor.

When copying PoolAllocators Bard experienced linking errors to the
copy constructor of Pool, which is deactivated. To fix this we added
the copy constructor. Note that it does not copy the pool such that
ownership of allocated memory chunks is clear.

Kudos go to Bard for noticing this and providing the fix.

In addition we removed the copy constructor of the specialization for
void as the default coy constructor suffices.

Its call was removed from dune_mpi.m4 back in 2012.

The warning with more then one job was: "make[1]: warning:
jobserver unavailable: using -j1. Add `+' to parent make rule."
Still, I get no speedup for -jN with N > 1.

Deprecate header and macro dune_static_assert.
Use static_assert instead.

They cased compiler warnings.

DUNE_FINAL is a helper macro for C++11 and should be placed in
std, where now all compatibility has to be placed.
Beautify Doxygen documentation.

sadly the touchpad added some arbitrary "pwd; ", which slipped my
test, as I don't use cmake. Thanks to Andi Buhr for pointing this out.

up to now you could specify a BUILDDIR variable, which implied that modules were built in
$srdir/$BUILDDIR.
Imagine you have your dune modules in $HOME/Src. When you set BUILDDIR=build.g++
your dune-common module is built in $HOME/Src/dune-common/build.g++
Now you change BUILDDIR to an absolute path, e.g. BUILDDIR=$HOME/Build.g++
With the latest change dunecontrol will now build dune-common in
$HOME/Build.g++/dune-common/

Thanks to Angar for bugging me :-)

This patch adds an AutoTools test that tests whether the compiler
supports the 'final' keyword.  It defines the DUNE_FINAL preprocessor
macro, which expands to 'final' if the keyword is supported, and
is empty otherwise.

kudos to Elias Pipping, for details see (FS #1166)
as the new Dune requirements are >= g++ 4.4 we are sure that initializer_list is available.