Added fully-fully noco code to potential generator and charge density part. Proper handling of spin dimension for density and potential still missing

NOTE: The usage of the alternative k point sets is not yet as planned. Actually
It was planned that multiple k point sets could be used in a single fleur run.
At the moment the input selects a single k point set from those provided.

...this might not yet work in all computation modes.

...dirty implementation to get the GW interface running

Bugfixes for Vacuum part in spin-spiral calculation. Still not 100% correct as conribution of last k+g that are only present for one spin are missing. Might not be relevant at all since these are expected to be tiny.

Modified sorting routine to use secondary key and used this in several places to simplify sorting of LAPW, stars...