Revisiting State Augmentation methods for Reinforcement Learning with Stochastic Delays