Tight binding 'hopping_stack' uses relative k-points, while BZ interpolator uses absolute k-points

[HugoStrand]

Dear Triqs developers,

This is a small observation that the tight binding class has the "old" behaviour assuming that k-space vectors are given in relative coordinates, while the "new" Brillouinzone mesh interpolators assumes that k-space vectors are in absolute units.

The offending tight binding function is hopping_stack among others, see
https://github.com/TRIQS/triqs/blob/3.1.x/c%2B%2B/triqs/lattice/tight_binding.cpp#L198

Best, Hugo