MCTS Deep Regret Minimization in a Multi-Agent Incomplete Information Game of “Take 5”