Contrast improvements issues: ensure that WCAG 3 APCA is used over WCAG 2 AA/AAA contrast rule

As outlined in https://github.com/jupyter/jupyter.github.io/issues/587, I find some changes made in good faith aiming to improve accessibility metrics, unfortunately do not achieve intended end goal of actually improving accessibility but result in the contrary. There are a few issues with changing colours, including:
- the relative importance of elements in the layout gets changed and may create a distractive environment (this requires to rethink how the layout should look like with the new colours!)
- the number of visually distinct colours decreases because we restrict luminance bringing remaining colours closer (especially affecting partially colourblind people)
- users prone to migraine might experience more exacerbations when working in high-contrast environment

While all are important (and were discussed previously in various issues and accessibility meetings), I want to draw attention to a new problem, which I was experiencing for some time now but did not understand on technical level and could not express it clearly; it comes down to the limitations of the tools which we use when evaluating contrast.

Most tools in 2021 use the WCAG 2 algorithm which computes luminance difference ignoring most aspects of human contrast perception. WCAG 3 will include a much better algorithm called _Advanced Perceptual Contrast Algorithm_ (APCA) which solves some of the issues.

APCA is already available in the Chrome DevTools 89 (see [Color and contrast accessibility](https://web.dev/color-and-contrast-accessibility/#advanced-perceptual-contrast-algorithm-(apca)), and [how to enable it](https://developer.chrome.com/blog/new-in-devtools-89/#apca)) and guidelines are included in the draft of [WCAG 3](https://www.w3.org/TR/wcag-3.0/#visual-contrast-of-text). A very colorful APCA contrast calculator is available on: https://www.myndex.com/SAPC/.

May I suggest that going forward any changes to colours aiming to increase contrast/accessibility will be evaluated with the use of APCA over AA/AAA rules from WCAG 2? At the minimum, could we consider both and discuss whether a change is really needed?

Ideally we would perform some user testing like an informal anonymous survey for each major change. I would not want it to be something with high burden (not even surveymonkey or google forms) - just a simple Twitter poll or equivalent.