State-Augmentation Transformations for Risk-Sensitive Markov Decision Processes