The dropsonde data suggested somewhat higher winds, while the onboard SFMR suggested lower winds (I don't think it showed anything above 123 kts). With that conflicting information, it seems like they just went with the standard 10% reduction from 700 mb for the final wind estimate.

The dropsonde data is not necessarily representative. As Random Chaos alluded to above, the dropsonde may report very high winds if it gets caught in some sort of meso-vortex which are not representative of the system as a whole. In one of the monster hurricanes of the last few years (I can't remember which off of the top of my head), I remember seeing a dropsonde wind observation of over 200 kts. The other dropsondes into the northern eyewall of Dean late last night (looks like there were 5 total) did not show anything higher than 136 kts in the lowest 150m.

Determining the maximum sustained winds of a hurricane is a very tricky business and probably something that is not done very accurately, even in this advanced age.