Core: mu_s scaling of LLG equation in effective field terms

[GPMueller]

It seems the implementation is missing a factor 1/mu_s in the LLG equation's effective field terms:

[GPMueller]

It may be a good idea to first move the mu_s factors from the Hamiltonians to the Geometry (see #364).

[GPMueller]

#364 has been implemented with 3e4df99, so this issue could be implemented now. Note the relation to #143: it would definitely make sense to have a list of mu_s values for all spins, as this would greatly ease the implementation of scaling for LLG.

[GPMueller]

With 978682a, mu_s and cell_mu_s are now distinguished. The former can now be used to correctly scale LLG. Might be worth adding a function void scale(vectorfield & vf, const scalarfield & sf); next to the already existing void scale(vectorfield & vf, const scalar & sc);

[GPMueller]

Fixed with 537b87c.