Thinking to Recall: How Reasoning Unlocks Parametric Knowledge in LLMs - Explained Simply | ArXiv Explained