ReasonOps Paper Reveals Hidden Structure in AI Reasoning Models