TODO: Depth to Heightmap

[giulioturrisi]

We could add the possibility to generate an heightmap from depth images, in order to use it for sim-to-real with only depth cameras. The pipeline could be

1. train the policy with heigmap (implemented)
2. in the play.py, implement a data collection routine to learn depth to heightmap via supervised learning

Status: Working on it, https://github.com/iit-DLSLab/basic-locomotion-isaaclab/tree/main/scripts/depth_to_heightmap

[giulioturrisi]

Done in main, need to be tested