Text this: Improving GNSS Positioning Correction Using Deep Reinforcement Learning with an Adaptive Reward Augmentation Method