A Goal-Oriented Adaptive Sampling Procedure For Projection-Based Reduced-Order Models With Hyperreduction