The initial and reported magnetic m_y moment is wrong in some subroutines

[Roberto Robles]

In noncollinear and spin-orbit calculations the convention for the m_y spin moment is not consistent through the code. Subroutines fill_dscf_from_atom_info (where the magnetic moment is initialized) and spnvec (used to calculate the magnetic moment) are not consistent with the convention used in the rest of the code. The subroutine moments.F calculates the correct magnetic moment.

In addition the actual initialization of the magnetization is not the one intended in the input file.

The bug will be present in any noncollinear or spin-orbit calculation where m_y=/0 (phi=/0). An example is attached where the behaviour is shown. This example proves which is the right convention.

Version: up to trunk_564 and 4.1

[Roberto Robles (roberto-robles)]

The bug has been addressed in revision 567 of Ramon Cuadrado's branch.