UneVEn: Universal Value Exploration for Multi-Agent Reinforcement Learning