Sample Efficient Myopic Exploration Through Multitask Reinforcement Learning with Diverse Tasks