First of all, I am not trying to argue that long term exposure to elevated BG levels is not damaging. I am familiar with both the DCCT and UKPDS data that show a nice correlation between HbA1c (a reliable proxy for average BG levels) and rates of microangiopathic and macroangiopathic complications. My argument is with the "damage starts to occur at levels over 140/7.8". which is widely quoted in the diabetosphere. I cannot find any published experimental evidence to support this idea of a "threshold" at 140/7.8 or indeed any other fixed BG level. As you say, the situation is more complex and most area under curve considerations are important.
Given that there does not appear to be ay experimental evidence to identify a specific threshold at which cell damage starts to occur it is interesting to ask why the figure of 140 is widely cited in the diabetosphere. I suspect that it is yet another example of the well known phenomenon where a "statement" that is made on the internet is re-quoted multiple times gaining apparent validity on the grounds of the frequency at which it is repeated.
In the end, if there is no scientific evidence for a specific threshold value, then it is not valid to quote ANY number as a definitive threshold! In fact, an average BG of 140 would predict an HbA1c of 6.5%. UKPDS data would predict that at this HbA1c there is indeed a statistically significant increased risk of complications, BUT at this level, the increase in risk as actually rather small. Of course, a person with an average BG of 140, would be predicted to be spending ~ half their time at levels above 140 and probably quite a bit of time well above 140 (ignoring the difference between median and mean for the moment). On that basis you might expect that if there is a damage threshold, it is likely well above 140 (otherwise the frequency of complications associated with 6.5% would be expected to be much higher).
Of course if anyone wants to set their CGM alarm to go off at 140, that is their own choice. I take the point about setting the warning at a lower level. The downside of that is alarm fatigue.