GitLab

I originally made the changes because the scale parameters acted differently
than in version 0.26. It turned out that the behavior in 0.26 is wrong, i.e.,
it does not comply with the documentation.

The scale parameters act on the x, y, and z axes. They do not act on the
lattice vectors.

It can be deactivated by using the command line switch -no_cdn_hdf

With -old mode only the old inp file is generated.
Without -old mode only the new inp.xml file is generated.

Also added some checks and errors if there already is an input file present.

In setab.f There was a problem that for some use cases "scale" had to
be index in a certain way while in others in a different way. I removed
these different cases by directly multiplying the elements of the lattice
vectors a1, a2, a3 in inpgen/lattice2.f with the a, b, c parameters. For
the cases when inpgen/lattice2.f is used the scale vector now is 1.0
after the routine.

- The automatic line breaks after ?? characters are now disabled.

- Relaxation of atom positions now works as it should.
- The code also becomes clearer.