Continual Reinforcement Learning on HumanoidBench (First Five Tasks Sequence Final Checkpoint)