docs: add ros-gazebo-gym positive reward comment

[rickstaa]

This commit adds a comment that lets people know that they have to set the positive_reward argument to True if they want to use the environments in the ros-gazebo-gym package with the stable RL agents in the stable-learning-control package.