Update RL examples: Drive and VehicleFollowing

[Adaickalavan]

Updated RL examples: Drive and VehicleFollowing

• RGB image fed into the neural network uses higher resolution and is cropped to focus on frontal view.
• If mission goal is present, the goal position is superimposed onto rgb image before being fed into the neural network.
• Refactored the RL training script.
• Training scenarios for Driving SMARTS 2023.1, 2023.2, and 2023.2 benchmarks are updated in the docs, in the benchmarks, and in the training examples.