Anisotropic - Your post struck a nerve. A bunch of them in fact. I have a DS who has dreamed of being a theoretical physicist since he was about 3, and has a school and testing history not unlike yours.
First thought: Elementary school math is typically all about computation, and tends to reward speedy calculations. It's not that uncommon to have strengths in conceptual math, but also weaknesses in retrieval of math facts. In other words, you can be really good at math, but still really bad at the stuff they call math in elementary school.
Second thought: While WISC testing is generally fairly reliable and stable by age eight, there are exceptions, too. Here's one such story; take it for the anecdata it is. Like you, DS was tested at 8 (just for giftedness). But he was also tested again at 10 (for writing LDs and anxiety) and 14 (LD and ASD). From virtually the moment he was born, he has always shown an incredibly affinity for math. In school, he had enormous problems with writing and, as he grew older, language expression more generally. Yet the testing at 8 showed high verbal, average fluid reasoning, overall mildly gifted. He was compliant in the testing, liked the tester, and she thought the results were reasonably accurate. And yet, there were some pretty major discrepancies with what we saw in real life.
When school problems cause us to do more testing a couple years later, I was both profoundly shocked and yet somehow also not at all surprised that DS's scores jumped up about an SD overall. We got a picture that was more in line with what we experienced in real life, but there were still some real oddities in the data, and overall results still suggested language strengths and math weakness, even though his real-life achievement was the opposite. After another four years of escalating school problems, more testing (and first time with the WISC-V, so we expected scores to drop), and most everything's gone up yet another SD or two, and for the first time fluid reasoning tests not just in the gifted range, but PG. Visual-spatial, on the other hand, has gone from being his super-power to his lowest score (except processing speed), because of the timing requirements. With that caveat, the testing finally presents a picture that feels rather more like this child, who has been notably unusual since literally the day he was born.
So what changed over those 6 years? I don't know. Did the earlier tester simply miss how much DS's anxiety and inattentive ADHD affected his willingness to give answers when he wasn't absolutely certain? Did he actually develop and grow that much, and the tests genuinely reflect where he was at, at that time? Did LDs and some ASD-like weaknesses get in the way of his ability to respond to certain kinds of questions? All I can say is, testing is pretty reliable, but it isn't infallible. When one-off, one-time, very specific measures don't seem to agree with your much more complex reality, trust yourself, and trust reality.
Because if I'm going to be totally honest here, I've read the same kind of studies you have, and have had the very same fears over the years. I've often wondered if DS was just surfing on high VS and working memory and was going to hit a mathematical wall, a hard limit on his conceptual capacity, at some point. Should I should be gently nudging him towards, well, almost anything other than a dream that only included theoretical math or physics? But math brings him joy, genuine, true joy. And I am so glad that at no point did I ever give room to that fearful voice that tried to whisper, "but what if at some point it gets too hard?" Such a day may or may never come. Live in this one. Believe in this kid that I see, and the evidence of my own eyes (and, hmmm, testing results that seem to be slowly catching up to what we felt like we already knew?!). In the meantime, he's learning, growing, expanding his horizons to know there are also many other choices out there in the world - but at this point, it's still the theoretical math that brings him joy. I hope it always will.
p.s. love your username
