Using approximate gradients in developing an interactive interior primal-dual multiobjective linear programming algorithm