Add Gymnasium RL interface and stable-baselines3 RL examples

[JacopoPan]

Roadmap

• allow parallel execution of multiple sim_run.sh (using different docker network subnets, headless/no GUI)
• Python-wrap the STIL environment in a gymnasium.Env class
• Determine a range of useful simulation steps (1Hz for high level actions, 50Hz for offboard references)
• Determine the observation space and sample it (e.g. using docker exec to echo ROS2 topics)
• Determine the action space and implement it (e.g. using docker exec to start ROS2 actions or publish offboard references topics)
• Add a reward signal with customizable implementation
• Create a stable-baselines3 example for a navigation task

[JacopoPan]

Closing temporarily to prioritize HITL