Dissecting Quantum Reinforcement Learning: A Systematic Evaluation of Key Components