Hybrid Deep RePReL: Integrating Relational Planning and Reinforcement Learning for Information Fusion