Constrained Hubbard

[mtaillefumier]

This PR implements the Hubbard calculations in SIRIUS using constrained occupancies. The non-colinear case is not covered in this PR (although it will work) because qe does not support it and we should find a way to describe complex numbers using the json format. The API has one additional call that set which orbital should be constrained.

[mtaillefumier]

I do not mind enforcing indentation but please merge the formatting PR first then I can fix all these needy greedy formatting errors.

[toxa81]

@mtaillefumier we should do the opposite - merge this first, then apply formatting

[mtaillefumier]

I still have two things to fix

• the order of the Ylm between sirius and qe
• q-e-sirius is buggy right now. Some problem with the interface

[mtaillefumier]

I think I fixed all remaining issues. pure QE is buggy with original implementation and only converges after 151 iterations while q-e-sirius converges after 31 iterations.

[toxa81]

@mtaillefumier ci/cd passes, so it's good from my side. OK to merge?

[mtaillefumier]

let's merge.