I've tried speeds between 6 and 300 rpm (equivalent, with my pulleys, to linear speeds of ~4 to 200 mm/s); the behavior is the same over that whole range. I tried reducing acceleration until I could see the motor accelerating over a significant fraction of a second; no change. I also tried reducing current until the motor skipped steps relatively easily when I loaded the motor by grabbing the pulley with my fingers (less than 100 mA), and then with my fingers put torque on the pulley in the direction of rotation (thinking that if the motor was skipping steps, pushing it along would reduce it) -- no change. Throughout most of my troubleshooting, I've kept the current setpoints the same on the 8825 and 8824 under test (e.g. adjusted Vref to 125 mV on the 8825 and 410 mV on the 8824, in both cases giving 250 mA), or set the current limit on the 8824 higher -- it's not a matter of the 8824 providing insufficient current.
Just for grins, I tried changing the microstepping from 1/32 to 1/4 on both drivers -- no change.
Just for grins, I tried changing the microstepping from 1/32 to 1/4 on both drivers -- no change.