I really think this question is more about the validity of progressive enhancement.
If you think progressive enhancement is a good thing, then it's also a good thing for accessibility.
If you think a site is not properly designed without progressive enhancement, then it's also not properly accessible.
Progressive enhancement 'fundamentalists' are in the minority, even in the accessibility community, but their arguments are strong, and worth considering, even if you end up compromising them
WCAG does not require or mandate javaScript, but it does require that any accessibility implementation is actually usable by real users in the real world. The jargon they use to describe this is "accessibly supported". I can imagine situations where an implementation requiring javaScript to do the most basic things (like navigate the site, or browse the content) is not sufficient to earn this label. This would be a problem.
I know of at least one accessibility-focused website which (for polemical reasons) sometimes requires javaScript to be disabled. The site has a video on progressive enhancement which might be informative.
And a heads-up: The Accessibility Object Model (AOM) is currently being developed by Google, Apple and Mozilla - this is is a javaScript API which will allow developers to manipulate the accessibility tree directly without messing around with markup attributes.
Accessibility features achieved via AOM alone will (by definition) not work with javaScript disabled. I expect this will change expectations about the value of progressive enhancement for accessibility.