preprocessor statements to remove that code

Note: The timings are still buggy:
   1. They are not yet reset in each iteration
   2. The timers should have other names.

Note: The output for valenceDensity, valenceCharges, and allElectronCharges is
      preliminary. It does not look nice in out.xml. The date should be
      collected in the routines to enable a nicer output at a different place.

1. The total forces are now written out to out.xml iff l_f=T.
2. There were bugs in some array dimensionings in force_a21. Corrected.
3. Some cleanup in cdnval, force_a21, and force_a21_lo.

systemwide, you have to compile it yourself and define a variable XML2LIB
to point to the static libxml2.a file.

...as suggested by Daniel.

The tlmplm arrays in that routine only have a spin dimension of 1.
Depending on the variable isp these arrays either contain the values
for spin 1 or 2.

Also in this commit: Enabled output of program version to out.xml.

Also in this commit: small bug fixes for core states output and
introduction of valenceDensity hierarchy layer in out.xml.

make doc will now generate html&latex documentation

Changed .eq. to .eqv. for comparison of logicals
Changed assignment of logical to integer variable

Made vdW stuff in vgen dependend on CMAKE variable Fleur_uses_vdW to avoid compilation if no fftw is present
Added CPP_MPI dependency in eig66_mpi and initParallelProcesses to enable compilation of serial versions
Made initParallelProcesses a F90 file

Note: This is incomplete.

Missing: 1. vacuum energy parameters
         2. alternative types of energy parameters