The trigger for MCAS to resume is that a pilot made a trim input. Otherwise it was a one-shot. So it only restarted in response to manual input. And it moved continuously over that finite subset interval of time. It did not change speed or direction.
So you agree that MCAS is not continuous, and therefore it is entirely reasonable not to characterise its operation as continuous/runaway, so not applying the runaway stabiliser QRH procedure is a reasonable (non) response?
The Runaway Stabiliser QRH characterises 'Runaway Stabiliser' as:
"Condition: Uncommanded stabilizer trim movement occurs continuously."
MCAS can be 'one-shot' or continual, interrupted by Stab Trim operations, and operating again if the MCAS trigger conditions still apply 5 seconds later. It is
not, as you point out, continuous, which is enough to put doubt into the mind of a cognitively overloaded pilot whether the Runaway Stab QRH is appropriate.
It doesn't help that STS operates by executing uncommanded stabiliser movements, so pilots regard that as normal operations. Seeing and hearing the trim wheels move is normal flight deck background, which is part of the reason why MCAS was so insidious.