Evidential Deep Learning for Proximal Policy Optimization in Deep Reinforcement Learning