Reinforcement Learning Optimizes Limit-Order Execution Using Logistic-Normal Allocation | Let's Data Science