Yeah, so we have two problems here: how the doubles get on there, and making sure they don't display as doubles when they do.
The tricky thing is that it's not clear I should stop them from displaying, because it's kind of just wallpapering over the problem. It should not be possible for you to double add, and I'm not sure it's realistic to expect you to convey the kind of granular detail needed to recreate how it happened in the first place.
So, absent that, I think it might be best for me to just delete the dupes manually (it's just that one list?), and basically ask you to be on alert for this issue when it recurs, at which point it'll be fresh in the mind and we can hopefully narrow things down enough to debug. And obviously, if at any point you find you can recreate the issue consistently with certain steps, we'll be in business, and a fix should (probably) be quite easy.
The thing is, the whole list appears as doubles, and my occasions for double tap have been much rarer. Could one accident cascade into a full duplicated list ?
And yes, it's the only list. But I'll try to be as clumsy as possible in the next one, to see if I can re-break something. Also, I intended to wait till tomorrow before trying to edit that list (swap titles around, delete some, and see how it affects display), I just wanted you to have a look at it first, and then to be sure I wasn't touching it
while you were trying to solve the issue. But maybe you already know whether my further experimenting with this already bugged list can bring more hints or not.
I have no real opinion about the double nature of the problem (origin + discreet self-correction) but I assume that, with complex programs, there are redundant failsafe checks that correct issues even if the issues shouldn't be possible. It's a matter of coding perfectionism, I suppose : it can feel as an imperfection to rely on safety nets instead of ideal codes. But I don't know how realistic ideal coding is, in the real world, or if, in practice, all programs rely on safety nets beyond a certain critical mass of complexity.
Anyway, it seems to be extremely rare (first such bug on more than 300 lists?), and it doesn't even affect the most important display mode. So, really no urgency here. And I'll be more attentive to the error message if it ever reappears.