Get Accurate depth map in meters from depth render

[hello-fazil]

The current depth render from the cameras is inaccurate and cannot be used for point cloud generations. Based on the pin hole camera model, we must implement a depth render pixel to meters function and also provide a pointcloud generation method.
Some of the API examples were mentioned in this thread:
https://stackoverflow.com/questions/66663291/convert-mujoco-depth-image-to-open3d-point-cloud