Enhancing Task Planning Efficiency of Reinforcement Learning Agents through LLM-based Action Masking | Synapse