Create README.md

comments + minor changes

added trained weights for 0.1 probabilities and 0.1 &0.3 probabilities

changed traffic generation to different probabilities at each direction

changed agent's parameters

changed baseline to support uneven probabilities scenario

removed constant max speed

project structure

added naive Round Robin algorithm for baseline

moved simulation initialization to Utils

minor changes

moved data outside env

added statistics at the end of simulation

added stronger seeding + documentation

added regular up-down traffic + changed model to the one in dqn atari article + changed Junction env implementation to non naive one