no worries kitz lets wait to see what sky say, here are my "theories"
I have learnt to not insist I am right.
Bear in mind attainable should be based on the target snrm so the modem reports what it believes attainable should be based on the targer snrm which is circa 6.3db. So during a power cut event the line still syncs at that snrm but when disturbers come online the actual snrm will drop to a lower number with the crosstalk but the attainable doesnt adjust for that. This explains why my attainable is not a high number.
I agree the fact I have no access to live stats during the sync, not only due to the dslstats not tallying at that exact moment but also that most of my network equipment would be disconnected following the power cut, I actually had no access to stats for quite a while due to my pfsense unit issue. However QLN is generated at the sync and doesnt update so I could still check the QLN data.
My line without any crosstalk should easily be able to get a 80mbit sync and should be able to get in excess of 100mbit as was the case when I first had FTTC, granted this is a different pair but this pair the characteristics actually are better than the original pair.
Now the 2 points you raised, the point of banding above what my line can achieve at 6.3db and the power levels?
The one purpose of banding the line the way they have done I expect is to prevent a line from syncing at a unstable rate following a power cut, the exact event that caused my line to sync so high, we know openreach have wide area event detections in place, so my theory is that somehow their systems can put it all together and determine that banding is beneficial, in which case I think DLM has actually done its job well, my line as it turned out did eventually have excessive errors at the 3db margin, after a second power cut its now synced at a more modest speed giving me around a 4.5db margin which should be more likely to not have the same problem. This is of course assuming I am banded.
With the ECI power level issue, my only theory is that if they are actually reducing power output (and its not a reporting bug) then they have determined it reduces errors to prevent DLM interleaving lines and as such is beneficial. Sync speed isnt everything, quality of service comes into play such as line stability and latency. If they have been reducing levels, then hopefully they will revert it after g.inp is working.