Update to align with stage 3 conditions

[rbuckton]

During the January TC39 meeting, the following changes are required to reach consensus for Stage 3:

• new.target should follow the same behavior as methods and field initializers and return undefined (Interaction with new.target? #25).
• A class may have multiple static {} blocks which are evaluated in document order interleaved with static field initializers (Why not allow multiple static blocks? #26).

In addition, I've updated the specification text to be a delta from the Static Class Features proposal.

In addition the following are required:

• Additional review by the ECMA-262 editors
• Additional review by @littledan

[rbuckton]

@littledan: You requested that I investigate leveraging Block rather than a custom parse node for ClassStaticBlock (i.e., `static` Block). I'm not sure this is feasible, as function bodies require additional semantics that aren't supported by the Static and Runtime semantics of Block, otherwise we would also just be reusing Block for productions like FunctionDeclaration, et. al.

I have, however, leveraged FunctionCreate and MakeMethod, and removed the custom class static block environment record in an effort to simplify the specification text.

[syg]

lgtm, I think all my comments thus far are editorial.

[ljharb]

LGTM pending shu's comments

[rbuckton]

@syg: I've made updates based on your suggestions. Please let me know if there's anything else you see that I should address.

@littledan: When you have some time, please provide your feedback.

[littledan]

This PR is perfect--it encodes the semantics we agreed on, and removes the duplication that I was most concerned about. Thanks for the update, and apologies for my delay.

[littledan]

Maybe this should link to the diff for now, so it is not a dead link?

[rbuckton]

Changing it to https://arai-a.github.io/ecma262-compare/?pr=1668 didn't fix the link. As soon as I open the diff it ignores the #sec-class-definitions-static-semantics-containsarguments added by ecmarkup.

Using https://arai-a.github.io/ecma262-compare/history/PR/1668/94b1af23d6ca075c6635bcd5b96c6450ec580f51 seems to work, though, but will get out of date if new commits are added.

[rbuckton]

I can kind of hack around it by using this biblio entry:

{
    "https://arai-a.github.io/ecma262-compare/?pr=1668&id=sec-class-definitions-static-semantics-containsarguments": [
        {"type": "clause", "id": "sec-class-definitions-static-semantics-containsarguments", ... }
    ]
}

Since that diff tool uses id=.

[rbuckton]

I updated the biblio entry above, and also generated a fresh ecma262biblio.json (would be nice if this were automatic) and fixed some algorithm references where the names have changed (FunctionCreate->OrdinaryFunctionCreate).

[syg]

lgtm!