d3q Code

Final state decomposition redefined in a symmetric form that preservs q2->q3 change for the coalescence part (the X part is unchanged, except for numerical noise which is potentially reduced)

Added the possibility to plot the final state in 3D via xcrysden using eitehr the xsf grid type or bxsf

Hand-made non-analytical contribution replaced with a call to nonanal, I thin kthis works but needs to be checked further as Simoncelli reported a case where LO-TO splitting with causes negative frequencies

Simple hand-made enforcement of sum rules on zeu removed as it is done by a subroutine further up

bugfix

possibility to spacify a plane of kpoints:

QPOINTS plane
n1 n2
e1x e1y e1z
e2x e2y e2z
q0x q0y q0z

cleanup

Optimization done with the highest smearing and highest temperature, which should be fool-proof

Optimized grids working properly in parallel, also done using normalized contribution of each band, which prevents loss of precision for bands with very low linewidth. I'm not sure it was worth, it works better with very small smearing.

Important note: the first value of smearing fro mthe list is used to impose the condition!

Avoid recomputing the frequencies inside isotope_scattering, this is required to have the correct energy when using LO-TO split along a path