Vorticity‐stream function approaches are inappropriate to solve the surface Navier‐Stokes equation on a torus