Besides, it doesn't matter whether it's a nuance, because apparently most engineers aren't competent in implementing accessibility. I'd rather that people who specialize in it can get paid to do it right instead of believe in a fantasy that engineers are going to going to care enough to do it themselves.
The problems come up when people see default <button> styling and then go off and make it a <div> instead. Nobody really teaches why this is a bad thing, you certainly won't find it in some random web tutorial on a blog from Google search. But if enough kludges like this happens in a code base it becomes pervasive tech debt in lots of tiny places, which is possibly the worst kind of tech debt.
I don't think it's impossible to educate engineers to do this, the problem is that we don't really educate about web development in the first place.
This really isn't a matter for developers to take on themselves. It is a matter of product development and feature selection.
So many people on here talking like they just "just do it right" and it's no more difficult to make the site accessible than it is to make the site at all. It is something which needs to planned an budgeted for.
It can get expensive. Especially when the development of the site is speculative and you don't know beforehand what it is going to look like in. Iterating over works in progress and maintaining accessibility along the way can be a real pain.
Hell, I can take the argument against this and apply it to things like QA. Why exactly can't engineers just do what QA would do themselves? Why not make all engineers devops? After all, shouldn't they understand the entire system from end to end? The answer is they can't, because the more skills you expect an individual to have, the more likely they're going to suck at everything they do.
The problem IMO is that a lot of people see the default styling of, say, a button, and go 'screw that' and make a div do the work of a button instead of trying to override browser button styling. A fair amount of people learn web development very informally, and in my experience very few tutorials or education sources actually talk about accessibility other than as something that sounds like something nice to have. So they don't know why doing this is bad, or they don't think it's a huge problem.
And it's very easy to do these small little accessibility kludges instead of modifying default behavior, and you don't realize what the cost of that is until lots of kludges later you get slapped with a lawsuit. If you're on a team prioritizing composability, this isn't necessarily very difficult to fix, but for websites that exist pre-composable frameworks and use a hodgepodge of older technologically, this is actually very difficult, especially since many of the people who wrote the original logic are probably gone and the code probably isn't commented sufficiently.
Accessibility is just another form of tech debt. I don't know of any places that would hire a tech debt specialist.