Reinforcement Learning and Evolutionary Equilibrium in Limit Order Markets