[gfs] Negative Matsubara frequencies and Pade

[krivenko]

Function pade() has an argument called n_points, which used to be understood as the number of Matsubara points taken into account.

With the introduction of the negative Matsubara half-axis, the meaning of n_points has to be clarified and the code must be updated accordingly.

Do we now need a pair of parameters, something like start and end?

As I remember, a similar issue was once raised about the tail fitting procedure.