Frequency Reuse sync source question

OK, the document "How to configure ePMP's Frequency Reuse Feature" tells us we need to handle things differently if the APs are on CMM4 or internal GPS, vs CMM micro.

Why?  Is there some inherent difference in the sync signal itself?  Also, if that's the case, which way do things go if one of the APs has a Packetflux SyncPipe as its sync source?