A Comparison of Hybrid Incremental Reuse Strategies for Reinforcement Learning in Genetic Programming