Gz 3D BBox messages are not bridged

[wittenator]

Desired behavior

There should be a bridge for the translation of 3D BBox messages coming from bounding box cameras.

Alternatives considered

None

Implementation suggestion

The corresponding gz messages should be added. I did this in #523 .

Additional context

This is useful if one wants to replace e.g. an object detection node that runs on CUDA-hardware only with a Gazebo bounding box camera to make the simulation run on any hardware.