Analyzing Explainer Robustness via Lipschitzness of Prediction Functions