[css-nesting-1][css-cascade-6] Clarify absolutization of the selectors of `@scope` and nested style rules

[cdoublev]

For nested style rules, the spec says:

When serializing a relative selector in a nested style rule, the selector must be absolutized, with the implied nesting selector inserted.

The related commit message refers to #8940, which seems unrelated. If I am not mistaken, it has to do with the implied specificity (#9069). Whether I am wrong or not, it might be helpful to explain the reason of the absolutization.

WPT covers this requirement in this test (non-exhaustive), which is interesting because it shows that :scope is not absolutized. As noted in #9621, in which it was suggested to absolutize it, it would change the specificity. But this issue is still open.

Chrome also absolutizes a scoping root selector when @scope is nested in a style rule. I cannot find where this is specified, nor a corresponding test on WPT. And it does not absolutize a scoping limit selector.

const sheet = new CSSStyleSheet
sheet.replaceSync('style { @scope (root) to (limit) {} }')
sheet.cssRules[0].cssText; // style { @scope (& root) to (limit) {} }

This somewhat suggests that <scope-start> produces <relative-selector-list>, at least when nested in a style rule. Should it?

Another question that comes to my mind is whether nested {} in style { @scope { nested {} } } is a nested style rule, a scoped style rule, or both. I think it is both, but :scope should still be implied (rather than &), as with any scoped style rule. Again, whether I am not wrong or not, I think it should be clarified.