Trust Region Policy Optimisation in Multi-Agent Reinforcement Learning