I guess, then, that we would also see a larger than usual claimed value between Oct. 30 and Oct. 31

Right! :-)

Now I include only Oct. 24, 25, 26 and 29, 30 and Nov. 1.
Although the days of both crashes do not match exactly (the first claim peak is a day earlier and the second on the same day), I still think these are the reasons. Perhaps the medians from the steemdb data have also shifted a little.

So I was thinking about it wrong

I don't know. So far, I have only identified a fraction of the possible changes to the funds in the code. So I can't yet say exactly how this is related.