Canonicalization for Calendar identifiers?

[anba]

Previously reported at tc39/proposal-temporal#2314, but that issue was closed after the case-normalisation part was fixed.

Q: Should new Temporal.Calendar("islamicc").id return "islamicc" or the canonical identifier "islamic-civil"?

If no canonicalisation should happen, it'd be good to mention this in https://tc39.es/proposal-temporal/#sec-calendar-types.

[justingrant]

Temporal meeting 2023-08-17:

• Per @sffc, we'll transfer this issue to the 402 repo. TG2 should decide whether a normative PR is needed.
• If we want to match time zone behavior, then DTF.resolvedOptions() should retain current behavior, but testing for equality (e.g. among Temporal.PlainDate instances with "islamicc" or the canonical identifier "islamic-civil") should canonicalize first before comparing for equality.
• There's also the cases of "gregorian" and "ethiopic-amete-alem" where those aliases are accepted in CLDR but they are resolved to the canonical IDs "gregory" and "ethioaa", respectively. Should behavior be consistent for these vs. "islamicc" ?
• UTS 35 explains the difference between deprecated and preferred (like "islamicc" → "islamic-civil") and alias (like "gregorian" and "ethiopic-amete-alem").

[anba]

• gregorian isn't a valid BCP47 identifier and therefore shouldn't be allowed in Temporal.Calendar.
• ethiopic-amete-alem is a special case -> Allow "ethiopic-amete-alem" as a special case? #285