d3q Code

Raise error when argument is expected but not present

Two changes:
1. grid shift can now be specified in input (xk0 and xk0_in), as a fraction of half the grid step
2. progress of long calculations improved, it now prints time and percent istead of the list of points

Added a small bash script that applies the sum rule directly to the fildyn files

Also, previous tools improved slightly

Better stopping on error

New debug option to write D3 matrices that do not match after imag part of FCs has been stripped

Possibility to do a calculation with nbnd_occ = nbnd, added as debug option

alignment fix

quick return of the ij term when the fermi shift is zero (i.e.e high symmetry crystal)

better code alignment

--This line, and thos below, will be ignored--

M incdrhoscf2.f90

Tentative try to get the same result with nbnd_occ = nbnd and without, also without breaking the code in the insulating case when nbnd>nelec/2

I'm relatively convinced that the correct nbnd_occ to use in solve linter is the one at ik_wfc (i.e. at k) and not at ik_prj (i.e. k-q in the simple case), but I'm not 100% sure. However, it does go much better like this

Also, it is still not perfect, and there are more residualse in the FFT when using this with respect to using nbnd_occ=nbnd