I don't image TR069 has anything to do with something as basic as the sync choices made by the modems.
It is obvious that, once synched, the modem calculates a "max attainable" speed as though that extra 0.5dB is available, so calculates something slightly higher than the current. At this stage, they are obviously acting as though 6dB is the target.
However, it looks like (for you) the modems are using a 6dB target during the sync, but then miscalculating the bit loading somewhat, and ending up with an actual margin of 6.5dB.
My guess is that the modems (unsure which end) are making the mistake at sync time, allowing for something within G.INP that isn't happening (at least, not yet). It might be a mistake, and it might be deliberate (and we just don't understand it enough yet). The current sync will allow for the (very small) bearer 1 to be carried - so perhaps the spare margin is something held back so that bearer 1 can expand if necessary.
Perhaps it is something used by SRA, or SOS. Does anyone know if either of these have been activated?