And further to poking things out of sunroofs...
A few years ago I compared the drag of two rear vision mirrors using this rig.
The approach was to move the vertical pole's attachment to the cross-arm closer to one mirror until, under the influence of the airflow, the pole no longer wanted to rotate ie the assembly no longer developed a torque. By looking at the ratio of the distance from the centre of each mirror to the vertical, the relative drag values could be calculated.
In this case, the Honda Insight mirror (left) had about 1.5 times the total (CdA) drag of the Yamaha R6 mirror.
By then measuring projected frontal area, the difference in Cd could also be calculated. This showed the Honda mirror to have a Cd about 20 per cent lower than the Yamaha mirror.
So, because of its shape and size, the Insight mirror had a greater total drag but a lower Cd.
You could also compare a mirror to an object having a known frontal area and Cd eg a flat plate.
Note that this test compares the drag of the mirrors in isolation - not when attached to the car.
A similar 'rotating axis' approach was used by the Wright Brothers to investigate the performance of aerofoils. They used a bicycle as the moving vehicle.