replace old model with new model compatible to the PoseEstimator, the old one will not work anymore.

adjust the output of the pose net to have more debugging data, like the po and transformed_depth image. Also adapts the input form of the pose model, from [[transformed_img, small_cam], transformed_depth] to [transformed_img, small_cam, transformed_depth]. Old models will not work anymore! Also adjusts its return.

outsource the calculation of the transformations of cam_K, depth and rgb image to functions in the utils.py and thus make it suitable for different sizes of rgb and depth image.

fix the centralization of the object within the bounding box and add the bbx_offset parameter to the convert_from_tfbbx to make the bounding box a little bigger. must be a number between 0-1.

fix camera resolution error, resolution of depth image and resulting po image is reduce, thus the camera matrix must be adapted to.