One minor snag
I'll go out on a flimsy limb and suggest that we should probably limit the set of possible variations to those hands in which a CVBJ user would actually be likely to split three times against a "ten". Therefore, the "Treys" variation isn't really valid, is it? ;)
Although it's probably not a possibility in CVBJ, I was thinking more along the lines of an error in a sub-routine responsible for DISPLAYING the most-recent picked card, not an error in the sub-routine that actually randomly picked the card in the first place. A Boolean which flagged the hand as reaching its maximum number of splits (SP3?) could prompt the jump from Eights to Queens, in theory. Of course, in such a scenario, an error like the former would be far more likely to display four Eights of the same suit, followed by four Queens of the same suit. There are all sorts of ways the hand could pop up without being a "valid" hand, but you're right: If it were a problem with CVBJ, it most likely would've been reported by now.
Although I agree that the chances of a goof of this nature are incredibly remote, the thrust of my post was that the chance might actually exist somewhere in the one-in-a-hundred-million range. That would make it far more likely than the actual "real" hand itself (or any of the similar hands you propose).
Still, I get your point. :)