Overcoming Brittleness in Pareto-Optimal Learning-Augmented Algorithms