It looks to me like the air speed in the tunnel is not sufficient to fully extend the tufts on the side of the bike.
I have seen that on my car at air speeds of around 20 mph. When air speeds are high enough the tuft follows the local flow direction.
If the flow is attached (good smooth airflow over the surface) the tuft will be pinned to the surface and arrow straight. If there is turbulent flow but still mostly attached the tuft will flutter over about a 15 degree arc centered on the average direction of air flow.
I set up my camera on rapid fire mode and shoot 5 - 10 shots in rapid succession so I can see the range of motion of the tufts. If you pick the right shutter speed so the fluttering tuft blurs a bit it is easy to see the pie shaped wedge it is fluttering in.
If the flow is totally detached and turbulent the tufts can point in almost any direction even back toward the direction of the air flow due to recirculation as the air tries to fill a low pressure area.
At low air speeds the tufts on the sides of the vehicle tend to droop below the true air flow direction and if you look at that bike all the tufts on the vertical surfaces are drooping downward.
I would suspect that picture was taken at low air speeds so as to demonstrate the method without giving away much useful info on the true airflow at speed.
See my posts on tuft testing on the other aero thread about aero stability
http://www.landracing.com/forum/index.php/topic,8252.210.htmlLarry