UNCLEAR: A Straightforward Method for Continual Reinforcement Learning