Added a new agent based on deep reinforcement learning

Created by: chriss1245

I added some new files in order to include the DeepQLearningAgent. It uses a very simple (from youtube) neurtal network instead of the qtable.

I have trained 3 different agents. The best is already selected.

Note: Most of the commits were issues with the log file. I had to do some tricks in order to be able to push. The thing is that the deep q learning agent required a lot of iterations to learn. The log file became greater than the maximun size allowed by github. Anyway. Think it works now.