Correct LiDAR (3DBox) -> Camera (3DBox) calculation

[Petros626]

I used the two original methods lidar_to_camera and lidar_to_camera_box to calculate a label in camera frame. I realised that the y-value never matched the original label, even though I was using the appropriate calibration file instead of the mean values in cnf..

Original camera label (h, w, l, x, y, z):  [1.48, 1.64, 3.59, -7.09, 1.75, 10.15, 1.57]
My approach output (x, y, z, h, w, l):     [-7.09, 1.75, 10.15, 1.48, 1.64, 3.59, 1.57]
Complex-YOLO output (x, y, z, h, w, l):    [-7.09, 1.01, 10.16, 1.48, 1.64, 3.59, 1.57]

Complex-YOLO uses the wrong matrix multiplication (order, no tranpose) and no height adjustment (reference point from box center to bottom)