Camera frame needs to be added programatically to any of the quadrupeds in the library

[Danfoa]

The current operation of the RGBD Camera instance depends on the existence of a camera body in the model/robot mujoco .xml description.

To be robot-agnostic we need to introduce the camera body programmatically before loading the .xml model file.

To do so we need a:

• Parent body name/id which must be a body of the robot e.g. head_body`.
• Homogenous transformation matrix of the camera frame w.r.t to parent frame.
• Add programmatically the camera xml element, before loading the mujoco model.

This is related to the PR #6 introducing the camera module with clean Python API, which is at the moment merged to devel branch. But this issue needs to be resolved before pushing to master.