Improve `create_magnetic_allotrope` function for the QE generator

[giovannipizzi]

The function is here

Two points:

1. The function should probably become a calcfunction, and get also the magnetization_per_site as an orm.List
2. The function changes the order of sites, sorting them by species name! We should as much as possible keep the same order, this makes things confusing for the user who passes a structure and gets a different structure with atoms swapped.