Xnr,
To answer you quite specifically, the autopilot might have a problem beyond its published limits, or it might not. You don't know until you've tried it, but you would know if it were close to a real limit, based on its behavior. If it starts getting slow and sloppy on the glideslope, and if you get to near zero torque (almost autorotation), then the autopilot is about to give up the ghost. Otherwise, most likely the limit is just a max demonstrated (sometimes the RFM is carefully worded as a "max demonstarted" limit).
We flew the DGPS decelerating approaches as full coupled and as flight directed, and only hand flew the ones that did not decelerate. When full coupled, if we entered at 80 knots or faster, and with 9 degrees approach angle, the approach resulted in near autorotation. Otherwise (slower than 75 knots, or 6 and 7.5 degrees) the approaches worked well.
Usually, it is easiest to set a limit right at the beginning, because the aircraft is already instrumented with the right measurements, and everyone is prepped, with the FAA set in "approve" position. It is a bit harder to get things spun up afterward, but still not a big thing, generally.