Modify shape in PredictedObject and TrackedObject

Discussed with @mitsudome-r and removed for several reasons.

• The DetectedObject does not have multiple shapes, so we want to standardize.
• Even if it can define multiple shapes, we can't offset the position of the bounding box because the shape is relative to the kinematics. !44 (closed)

For these reasons, we decided to use only one shape, and we think carefully about articulated vehicles again!

@mitsudome-r If you have any additional information, please let me know.