Update eigenso.F90: For Wannierfunktions with SOC but jspins=1 we need to make sure that we get also the spin-down component of the spinor. Therefore, the wannierspin was introduced, which is 2 when jspins=1, SOC=true and wannier=true.
For the results%neig and results%eig it does not seem to be necessary to write down the second spin when jspins=1.
Therefore, I changed this. This also avoids running into the MPI error which occurred when sending the second spin to an
array that has only memory allocation for one spin.