Reinforcement Learning for the Traveling Salesman Problem: Performance Comparison of Three Algorithms