On power and complementarity of the experimental constraints on seesaw models