Fin-R1: A Large Language Model for Financial Reasoning through Reinforcement Learning