Starting magnetization and constraints

giannozzrequested to merge
giannozz/q-e:startingmag into develop

Ensure that starting_magnetization does not retain funny default values when constraints are set.

Note: There seems to be an inconsistency between the acceptable values for the unconstrained and constrained "atomic" case. Not sure what starting_magnetization is supposed to be in the latter case (in the former, the percentage of spin polarization)

Merge request reports