R1-T1: Fully Incentivizing Translation Capability in LLMs via Reasoning Learning