- add readme and images for layouts

[vavalm]

• complete main readme
• adapt readme for docker containerization
• add Dockerfile to create the image based on official tensorflow docker image
• add requirements.txt
• modify run.sh
• modify pacmanDQN_Agents to avoid exception if we put not a String in "save_file" parameter

• save more regularly
• let a save example

[vavalm]

Hello I'm a students who work in Dqn algorithm.
I have a project in which I use Dqn on Pacman and I saw your repository that was very helpful for me.
I propose a pull request with the listed modifications.
One of the main add is the portability with Docker based on the official tensorflow docker image. The container is full ready for a GPU use.