I think one point of the OP question has been not addressed in the responses so far, but I am not sure I know the answer to this (paraphrased) question...
"IS is able to effectively neutralize the motion of the lens+camera, to reduce photographer-induced motion blur in the photo. But it would seem that a suitably high shutter speed also neutralizes motion of the lens+camera by itself, too (which is why the 1/(FL*1.6) rule of thumb for APS-C cameras and 1/FL for FF cameras...so at what shutter speed is our photographer-motion sufficiently stopped by shutter speed alone, and IS becomes superfluous?"
...because the OP comment, " There are also times when some motion blur may be desired... " applies primarily to SUBJECT motion. Yet it could also apply to deliberate camera+lens motion while panning the camera to almost freeze the subject but blur the background in motion.
Upon further thought, it might be based upon the concept that, without IS at some speed above 1/(FL*1.6) motion blur of the lens+camera is so small as to be not noticeable, which I would conceptually identify as 1/K(1/FL*1.6). The answer is dependent upon the angular rate of change of the lens front as it swings thru space due to photographer shake...IS reduces the angular rate of change of the optics in the opposite direction. The value K might by hypothetically K=2 or K=3, or some other value related to the angular rate of change. Therefore,
θ = 1/(1/FL*1.6*K)
...,where θ is the angular rate of change.
Our problem is that Canon gives us no parametrically defined value for quantifying what IS itself is capable of neutralizing, so our K value is a total guess for trying to quantify it. Furthermore, the real answer to K and θ are both also related to the ability of the human eye to detect motion blur, which has some relationship also to the Circle of Confusion blur circle size that we can see, which itself is related to the size of print and the viewing distance that it is viewed from!
Any high school scientist out there want a topic for a paper?!