Task-Agnostic Continual Reinforcement Learning: Gaining Insights and Overcoming Challenges