CPPO: Accelerating the Training of Group Relative Policy Optimization-Based Reasoning Models