Add feature specification of static extensions

[eernstg]

This PR adds a proposed feature specification of the 'static extensions' mechanism. For background, see #723. The proposal is mainly an excerpt from an earlier proposal about implicit construction (here).

[eernstg]

Hi @stereotype441, I've extracted a proposal about extensions that are capable of adding static members and factory constructors to existing declarations that are where such members can be declared (classes, extension types, etc.).

I'd like to use an approach which is similar to what @munificent has done a few times for documents in 'working': The review would just be a brief sanity check, and then we'd handle all the discussions about details in separate issues here in the language repo.

There are two versions. One version introduces a new kind of declaration (static extension). I tend to prefer this, because the way the declaration uses the on-type in connection with constructors is so different from the way we're using the on-type of a normal extension with instance members. It is also an anomaly that we would need to allow the on-type to be an extension, because they can have static members, so we may want to add them.

The other version is a generalization of the extension declaration that we have today. This version has some other benefits: A rather large handful of things need not be specified in this variant, because we already have extension declarations, and they already have the rules we need.

Anyway, I expect that we'll have some really nice dog fights about this choice.

Could you take a look in order to have that brief sanity check?

[eernstg]

Several changes introduced based on the review.

[eernstg]

@lrhn, PTAL. I created an issue #4049 in order to discuss the main remaining language design question.

This PR does not update working/0723-static-extensions/feature-specification-variant1.md, it only changes variant2.

[eernstg]

Thanks for the input on working/0723-static-extensions/feature-specification-variant1.md!

I actually hadn't planned to work on that document in this PR (because that document is nearly a copy-paste of https://github.com/dart-lang/language/blob/main/working/0107%20-%20implicit-constructors/feature-specification.md), but I'll go over it now. This means introducing into 'variant1' many of the improvements to 'variant2' which were suggested in reviews on this PR, because they are similar in so many ways. This might preempt some comments on 'variant1', but I'll try to keep track of it such that nothing in the comments is ignored.

[eernstg]

NB: At this time, working/0723-static-extensions/feature-specification-variant2.md is in a reasonably good shape whereas working/0723-static-extensions/feature-specification-variant1.md needs to be revised (in ways that are similar to the changes made to 'variant2' recently).

[eernstg]

I made some changes to '...variant1.md' in response to review comments.

Note that I still need to go over '...variant1.md' and make a lot of changes that are similar to the ones that I performed on '...variant2.md'.

[eernstg]

Hi @leafpetersen and @lrhn, cc-hi @dart-lang/language-team. I'd like to land this now. I think it's better to take the remaining discussions via Github issues, and I also think the relevant issues have already been created. Then we can change these documents as needed later on, when the issues have been processed. WDYT?

[leafpetersen]

I think we're at a point where the most productive thing is to land this as an interim state and move any remaining discussion to issues. LGTM to land as WIP.