Online Sub-Sampling for Reinforcement Learning with General Function Approximation