If you look very closely, the line isnt perfectly straight.
ie
20:33 - 7.8
20.34 - 7.9
20:35 - 7.8
20:36 - 7.9
20:37 - 8.0
and so on.
The disconnection @ 5:01 was the DLM removing interleaving & INP from the upstream on bearer 1
At 21:40 you had a small sharp burst of downstream CRC errors at 21:40 the line disconnected. In the same period you had 11 SES (severely errored seconds). So it could possibly be a noise burst that took the line down. If it was just CRC I'd say it was normal CRCs that can occur when a line comes back up
Could some1 else's router have disconnected then they rebooted - caused crosstalk that disconnected my internet?
No. Crosstalk shows differently in your SNRm
In view of the fact you got some SES, far more likely to be an EMI noise spike - ie an electrical type source, for example someone flicking a light switch.
There is also a possibility it could be RFI, bearing in mind when I first looked at your QLN I saw possible signs of either RFI or EMI in the lower tones.
I find your QLN more interesting. There's some lovely signs of crosstalk going on... plus some RFI.
would the trenches & bumps not be caused by crosstalk?
No, that is not crosstalk. It shows differently and is obvious to the trained eye.
There is no correlation with the downstream, where if anything it usually shows the most. Your downstream seems to be displaying signs of daily variance swing. which is typical of longer lines. If you look at the general curve over the course of the day, it starts to get worse as night falls and is at its best during daylight hours.
Crosstalk is a constant that is there or it isnt. EMI/RFI fluctuates.