Dipole acceleration is wrong for mGGAs and hybrid functionals

For mGGAs, treated within generalized Kohn-Sham, and for hybrid functionals/Hartree-Fock, the dipole acceleration is not correct, as we are missing the contribution from the nonlocal operator.

This needs to be forbidden and maybe implemented.

See Eq. 54-57 in https://arxiv.org/pdf/2302.00862.pdf This needs to be carefully checked and possibly adapted before been implemented.