More spec ports, and small updates to GetTemporalUnit typings.

[12wrigja]

Like other spec port PR's, I will merge this as-is without squashing to retain history.

[12wrigja]

FYI that I'll be on vacation till the 6th of September (2022), so I won't respond till then. There's no rush on this review.

I've asked some colleagues to take a look at some of the more complex type-system implementation I used for PrepareTemporalFields to see if there's any way to simplify it further whilst retaining the useful properties the current impl has:

• preparing fields returns an object with all the given field names as required properties with the types from the "potential owners" of that field.
• all the other properties of the potential owners that weren't specificed in the field name list are marked as optional - this allows them to be assigned to without casting / complain from tsc. This isn't perfect - it does mean in some cases casts are still required to change the properties type after assignment (e.g. ecmascript.ts L4574 is a good example).

[shicks]

See also the final state of the playground I was working with while experimenting. I have a few links to earlier versions of it throughout the review (as well as a few other links to separate playgrounds, so there's that).

[12wrigja]

@justingrant @ptomato I think Steve and I are done with back-and-forth on this, and I'd appreciate if you could take a look. Thanks!

[ptomato]

LGTM although I didn't deep-dive into all the changes to the typing of PrepareTemporalFields.

Separately, if it turns out that the types of PrepareTemporalFields are that convoluted, it may be worth deviating from the spec text there (in a way that's not observable from JS, of course) and having separate operations for PreparePlainDateFields, etc. if their types are easier to express. But that would be out of scope for this PR.