Behavioral Evaluation of Hanabi Rainbow DQN Agents and Rule-based Agents