Translational invariant of the position matrix

[Chengcheng-Xiao]

As described and fixed by jaemol, the original formula used in W90 to calculate the Wannier position matrix doesn't follow translational invariant condition.

He's fixed this in his version of Wannier-Berri: https://github.com/jaemolihm/wannier-berri/tree/transl_inv.

Try use this and see if WOPP can be improved.