"Race is a socially constructed category" is a semantic argument, not a real debate point. An ethnicity indicates you are descended from some set of people, and share genetics and/or culture with them. A race is a group of people whose membership is almost entirely arbitrary. Just as an example, many (maybe most?) people who are Hispanic have some amount of Spanish heritage (the country, not the language). Spanish people are white in terms of race, though. It's an arbitrary boundary we've drawn. Ethnicities are actually tied to geneology, so they could potentially be useful.
I still get your point, though.
I think you're right in that abilities and disabilities can be genetic. It would make sense that groups that share genes could share abilities and disabilities. There are 2 things to note, however.
The first is that at this point in humanity's progress, I doubt that there are very many people who only belong to a single ethnicity. And each time people of different ethnicities procreate, each gene is effectively randomly taken from one of the two ethnicities. The chances of inheriting an entire multi-gene sequence from one parent is fairly low. That's why many of these genetic disorders are only common in communities that practice endogamy. For the wider population, unless you're descended fairly recently from someone that was a part of those communities, your chances are extremely low. I would expect that same here. If some ethnicity has an ability/disability, it would be common among an endogamous group of that ethnicity, but the chances for someone who split from that group several generations ago to have it is going to rapidly approach 0. One would expect that given random interbreeding between ethnicities, everyone would approach an average.
The second is that human behavior is extremely complex. Finding the causality for a delta in some ability is very hard to do. Most of the things we have traced back to genetics are binary. You have a disease, or you don't. There is no standard deviation on whether you have the disease or not, you either do or don't. Genetic diseases are also effectively immune to social factors. It doesn't matter how rich or poor, or white collar or blue collar you are; you can't get Tay-Sachs because you were discriminated against. Those add up to make it very hard to find the causality behind much of what we consider an ability. If one group ranks 10% higher in IQ tests, it might be because they're wealthy and everyone else is experiencing food shortages. Or they can afford better schools. Likewise, if another group can run 10% faster on average, it might be because they tend to work blue collar jobs and are in better shape than those of us that sit at a desk all day.
There are tons of factors that can affect physical and mental abilities. Genetics are likely one of those. If we can't even establish the causality, we certainly aren't going to be able to gauge the magnitude. Is it a drastic increase/decrease in ability, or is it a small change that is exacerbated by everyone overestimating it? It can become a self-fulfilling prophecy. If we demonstrate some ethnicity can run faster, they'll start getting picked more to be professional athletes. Then more members of the ethnicity will attempt to become professional athletes, thus raising the average running speed and making the delta look more significant than it is.
Overall, I agree that genetics probably plays a role. I just think tracing it to an ethnicity is a) too difficult to be realistic, and b) not targeted enough to be useful. The cost of genetic testing will eventually fall enough to where we don't have to look at ethnicity to guess at what genes you have, we can just test everyone. Then ethnicity is irrelevant, it's entirely about whether you have the gene or not.