The ES's are caused by CRC's, so take a look at that graph... You will see the big burst yesterday, when SNRM was 0. However, you will also see lots of bursts that are semi-regular: around once per hour (but only roughly), and for around 10-15 minutes.
The ES graph only samples once per hour, so you don't see the same granularity as you do on the CRC graph (once per minute).
Figuring out what happens with that kind of pattern will probably help solve the root cause.
But, look more closely at the CRC graph, and you will see a general "fuzziness" down near the X-axis, between all the obvious peaks. Those CRC's keep happening constantly, at a low level, but they are almost hidden by the height of the peak.
Think they've banded me 
From the speed & SNRM values, it does look that way. But any DLM intervention (if that is what it was) hasn't changed the INP or delay values - strange.
Equally strange is that, since the resync, you are getting a lot of downstream FECs. By hovering over the "XTR" box at the top, I can see (from "R=6") that FEC has been activated downstream without Interleaving (as that graph still shows a constant D=1).
So, even though you are getting a lot of FECs since the resync, and the
height of the CRC peaks has decreased, you are still getting
some CRCs at a fairly constant rate - so the slope of the ES graph hasn't changed.
If DLM did intervene, it did so in a funny way (by banding alone) that allowed FEC's to be turned on (but less effectively, without the interleaving), and it has had little outcome on the real issue - all those ES's. It would be interesting to see if there is any packet loss, and if that has changed from earlier.
and although there is still time, it does seem weird that you still appear to be on open profile. Openreach are obviously messing with g.inp and there is something strange going on, but no-one seems to be telling much.
It's almost like they've broken it, since pausing the ECI rollout...
The really odd thing is that I would have expected DLM to take action after yesterday
[...]
One thing I would suggest is that you said you had a power cut yesterday.
If the power cut was widespread enough, perhaps DLM discounted the day as a wide-area event.
Just a view of my traffic lights on DSLWS.
Do I have the record for error seconds?
An enviable record indeed!
