Optimizing Reinforcement Learning: Fog and Edge Resource Management Through Bootstrapping and Reward Shaping