Asymptotic analysis of a class of nonlinear penalty methods for constrained multiobjective optimization