The question Why doesn't ISS start to spin if people walk inside? is a good one partly because it has attracted several good, informative answers. One of them is particularly intriguing because it contains the following statement:
In the Mission Control Center, it was possible to tell if the astronauts were awake, based on how the CMGs were operating (four of them). But the motions are very small.
The physics makes sense of course. As astronauts move (especially) along the forward/backward (roll) axis of the ISS, the movement of mass relative to the center can change the moment of inertia about the left/right (pitch) axis. Assuming angular momentum is conserved this will change the rotation rate about the pitch axis.
As pointed out in this helpful answer the ISS rotates once around its pitch axis every 93 minutes in order to keep the "bottom" facing towards the nadir.
So if everyone "ran" to the front and back ends and waited a while, the rotation rate would slow and the ISS would slowly pitch-up. If they all then "ran" to the center, this drift would reverse and it would pitch down again.
My question is, roughly how sensitive is the attitude control system to this kind of activity? Is it something that can be detected over a period of seconds, or minutes, or more like a day? How well (and how rapidly) can the ISS attitude control system detect and resolve periods of astronaut activity?