The problem was that some arrays had predefined sizes that were
not adapted to the actual problem. I changed this for some arrays
now but there are still some other arrays that may become a problem
at some point, too.

This broadcast was missing for the use of this variable in vgen/vmtxcg.F90, line 292

This might have caused an error in mpi parallel mode.

This might have produced real bugs in mpi parallel mode.

At the moment this does not seem to be important as the final result
of vgen seems to be originating from irank 0 only. However, valgrind
complained that some jump in vgen->vmts->ingr (line 191) is undefined
because an initialization is missing for all iranks != 0.

This is activated by using the inpgen command line switch "-electronConfig"

Note: The temporary usage of the optional mpi argument in stepf still has to be removed.

This is one more step towards using the parallelized calculation of the
stepfunction in the inp.xml code path.

This is the first step to enable the parallelization for parts of this code
as it is provided by Uliana.

The array rme had 8 dimensions. This is only allowed by compilers that implement this
Fortran 2008 feature. Other compilers only allow 7 dimensions. I splitted this
array into the three arrays rmeA, rmeB, and rmeC, which all have 7 dimensions. I hope
this fixes the problem.

I adapted Roman's code slightly such that it fits into the new fleur version.
Nothing is tested yet, so testing and polishing still needs to be done.