You're pointing at less than 6% difference from the expected, with a random number generator - the chances of getting results like this are fairly high. I suspect the three seeds don't use different formulas so this is almost certainly just random variation.
I keep very careful count of the high level seeds every time I wipe. Since there has always been a nice end-of-a-bell-curve distribution in my experience, I think the odds of the bell curve having a bump at the end are not very high.