Also test:

- double precision without mpi
- single precision without mpi

In ELPA_GPU branch, also test single-precision builds

Setup a build test for ELPA without MPI

It turned out that if a CPU supports SSE the already existing
test for SSE assembly instructions always passes.
However, the compilation of gcc SSE intrinic instructions might
nevertheless fail if gcc is not called with one of the options
"-msse3", "-msse4" , "-msse4.1", "-msse4.2", "-mavx", or "-mavx2"!

Obviously gcc does still not consider SSE as a standard on X86_64
Intel CPUs.

An additional configure test has been introduced, which test for
gcc intrinsic sse instructions. If this test fails, the corresponding
kernels are switched off.

The C++ kernels can be written as C kernels, which simplifies the build
procedure

In order to increase type safty all ELPA2 kernels provide now
an interface. The interfaces for the C/C++ kernels are automatically
generated during the configure step

The utility binary printed the available kernels to stderr.
This is changed. The ELPA library itself still does all prints
on stderr

The test programs are just needed at the build step (make check),
they are useless for users and will not be installed anymore

Quite likely the FMA4 (AMD) implementation has never been tested.
There is a fishy intrinsic call, which is most likely an typo
Abort with #error at compile time

For the Intel compiler, this was assured with the pragma

  !DEC$ ATTRIBUTES ALIGN: 64:: a

however, other compilers such as gcc of course did not honour this,
which could result in SIGSEGVs in case the variable was not aligned to
32 bytes (by chance!).

This fixes issue #11, thanks to Nico Holmberg for reporting this.