IP absorption scaling factor still wrong wrt yambo output

[palful]

There is still a scaling factor missing in the IP spectra (real and imaginary parts) from ip_eps2 from YamboDipolesDB.

So far we are using:

cofactor = spin_deg*4.*np.pi/self.lattice.nkpoints#/self.lattice.rlat_vol

The missing factor is system-dependent, but not k-dependent it seems.