Predicting Hydrological Drought Alert Levels Using Supervised Machine-Learning Classifiers