Towards General Negotiation Strategies with End-to-End Reinforcement Learning