Add spec URLs for CSS Flexbox

[foolip]

Extracted from w3c@4f66425

[foolip]

I created this to do self-review of some changes and help answer #6765 (comment).

[Elchi3]

As said previously, I don't think we should link to snapshots / leveled specs. More suggestions inline.

[Elchi3]

An overall question in this PR seems to be:

Should we link to sections or directly to the definitions? There are suggestions going on both directions now.

Definition: https://drafts.csswg.org/css-box/#propdef-margin-left
Section: https://drafts.csswg.org/css-box/#margin-physical

@sideshowbarker for spec tooling, which one is better?

[sideshowbarker]

An overall question in this PR seems to be:

Should we link to sections or directly to the definitions? There are suggestions going on both directions now.

Definition: drafts.csswg.org/css-box/#propdef-margin-left
Section: drafts.csswg.org/css-box/#margin-physical

@sideshowbarker for spec tooling, which one is better?

I guess the question is, which provides a better user experience for web developers? When web developers are reading the spec, is it a better user experience for them to find MDN/BCD annos next to the section headings, or further down next to where the definitions are in the prose?

Personally, having the section headings be where the developers find the MDN/BCD annos seems like better user experience in the general case. However, it can been poorer user experience if we have multiple annos associated with a particular leading. But if it’s only single anno, then it seems better next to heading.

But that’s just my own subjective perception of it. The spec tooling itself is otherwise agnostic to it.

[Elchi3]

Okay, I think you have more experience with browsing specs and the user experience thereof.

If from a spec tooling and data perspective, both, sections and definitions are equal, then I would say let's take the user experience aspect into account and link to sections instead of definitions.

Other opinions? @foolip?

[foolip]

I prefer linking to sections myself since it will then also include any introductory bits before the definition.

There are cases where this won't work out great, such as when linking the members of IDL interfaces. If we link them all to the section which defines them, it's going to look repetitive and redundant.

But perhaps we can prefer sections for now and figure out that later.

[Elchi3]

There will likely be exceptions but lets prefer sections when possible then.

Can you make changes throughout this PR and then re-request my review?

[foolip]

@Elchi3 I'm finished updating things now :)

[Elchi3]

Just one nit per the CI failure. Happy to merge otherwise.

[Elchi3]

Thanks for sorting this all out, Philip and Mike 🎉