Bugfix to the Ewald2D for large slabs.

See merge request !714

SG15 pseudo set should not be marked as experimental. This had already been done in 076714cdut was unintentionally reverted, probably during a merge conflict originating from b9dd0f66.

further bug fixes relating to Si problem.

See merge request !694

Bug fix: some norms were badly handled in the implicit RK propagators.

See merge request !665

Hotfix magnus

See merge request !655

Fix for parsing some non-standard(?) UPF2 and QSO files

See merge request !635

Rename family_is_mgga variable to family_is_mgga_with_exc in propagator_init for consistency, as that's what is actually passed to the routine.

The terms were incorrect and the code assumed that when calling hamiltonian_apply the result was added to hpsi.

With these changes I'm able to obtain a stable propagation with the Magnus propagator.

The interpolation in the Magnus propagator is to compute the time dependent terms of the Kohn-Sham potential, no to update the hamiltonian! This bug was introduce in 087deac5.

Workaround for a problem with ifort 2018.1 and OpenMP.

See merge request !637

The Ewald long-range for 2D was leading to NaN for energy or forces for thick slabs.

See merge request !632

Hotfix unoccc output

See merge request !628

The variable OutputDuringSCF was not used for unocc calculations, unlike what the documentation was saying.