Normative: Revert to previous behaviour by setting fallback value for fractionalDigits to *undefined*

[ben-allen]

Fix #149 Fix #144

[sffc]

Two issues:

1. Please note in "Properties of Intl.DurationFormat Instances" that the field can be undefined.
2. We should pass 0/9 into Intl.NumberFormat instead of undefined/undefined in PartitionDurationFormatPattern

[anba]

Intl.DurationFormat.prototype.resolvedOptions currently asserts that all internal slots have a non-undefined value. This is no longer true if [[FractionalDigits]] can be undefined.

I guess an undefined [[FractionalDigits]] should be omitted from the output for consistency with the other resolvedOptions methods, even though undefined for [[FractionalDigits]] has a different meaning when compared to undefined internal slots for the other Intl service constructors.

[anba]

I also wonder if the entries in Resolved Options of DurationFormat Instances should be reordered if an undefined [[FractionalDigits]] is omitted, so that the optional [[FractionalDigits]] appears last. For example by moving [[NumberingSystem]] to the top, see also Resolved Options of DateTimeFormat Instances and Resolved Options of NumberFormat Instances.

[ryzokuken]

There was also the suggestion of reordering the resolvedOptions, but I think it's something we can build consensus on this evening.

[sffc]

TG2 discussion: https://github.com/tc39/ecma402/blob/master/meetings/notes-2023-06-29.md#revert-to-previous-behaviour-by-setting-fallback-value-for-fractionaldigits-to-undefined

[FrankYFTang]

This PR reach TC39 consensus in July 12, 2023

[ryzokuken]

Thanks for the update @FrankYFTang! I'll rebase and merge later this week.

[FrankYFTang]

Could we add some test to test262 to validate the change of this PR?

[FrankYFTang]

@sffc @ben-allen @ryzokuken
I do not belive the part of reordering the order of resolvedOptions to move "numberingSystem" to the top reach any consensus. The title of this PR has nothing to do with that and somehow that reordering to merged under this PR without TG2 discussion nor TG1 consensus. Please revert that change.

@ben-allen this PR is already merged to the trunk, you merge your restoration into your branch of this PR after the merging which will not fix the trunk.

The landing of this PR seems very very troublesome . it include part which we didn't discuss

The landing of this PR includes 4 commits according to
https://github.com/tc39/proposal-intl-duration-format/pull/150/commits

df63a06
4b9a46a
e4774b0
I believe these 3 are ok

32950ea
this is a troublesome change, it chagne two thing- it perform some necessary change for the above 3 commits but ALSO include the reordering change of numberingSystem in the resolvedOption
this later was never communicate and agree upon TG2 members and never mentioned to TG1 members.

I beleve our edtior should revert that part of the change (the order of numberingSystem in resolvedOption should not be on the top ASAP since that was a technical error for merging something we never agree upon. If we in deed want to reoder that. That need a Normative PR to be discuss in TG2 and also TG1.