Act Only When It Pays: Efficient Reinforcement Learning for LLM Reasoning via Selective Rollouts - View it on GitHub