MapExRL: Human-Inspired Indoor Exploration with Predicted Environment Context and Reinforcement Learning