Using Bound Sets in Multiobjective Optimization: Application to the Biobjective Binary Knapsack Problem