[AeroModder]

My guess would be that the boundary layer is thick enough to not move the tufts much at all, but not quite enough to cause separation with reversal. For example, if you look at the smoke testing for the Insight, you'll see the gap between the surface and the smoke widen.

There's also the spoiler to consider. It's creating an area of slightly increased pressure around the bottom of the window, but it's not enough of an angle to cause a locked vortex. I'd be interested to see a tuft test without it in place, if that was possible.