Add 3D Coordinate Information to Object Tracker When Depth Is Enabled

[Luxonis-Brandon]

Start with the why:

Our object tracker now currently returns the 2D bounding box of the tracked object. The metadata of the neural network returns the 2D bounding box, and also the 3D position (in x,y,z, in meters). So it is possible to get the 3D position of a tracked object, but it requires some additional (and mostly unnecessary) intersection-over-union (IOU) work on the host to do the association.

Move to the how:

Calculate the depth of the tracked object on DepthAI return the 3D position of the tracked object when depth is enabled (i.e. when -dd is not given), just like we do now for detected objects (by the neural network).

Move to the what:

Provided 3D XYZ position (in meters) of tracked objects when the object tracker stream is enabled and when depth is enabled (i.e. when -dd is not used).

[Luxonis-Brandon]

Since Gen2 does not have an object localization capability yet (i.e. object detection with 3D XYZ information), and does not yet have an object tracker, we will need to implement this node/capability first, and then pair with object tracker.

[Luxonis-Brandon]

This is now initially implemented:
C++: luxonis/depthai-core#92
Python: luxonis/depthai-python#173

[Luxonis-Brandon]

Code sample for this here: https://docs.luxonis.com/projects/api/en/latest/samples/spatial_object_tracker/