Align Duration.prototype.round() to latest specification

[HalidOdat]

Resolves #188

[HalidOdat]

Running the test262 with boa with temporal main vs PR we get one plus passing test :)

Test262 conformance changes:
| Test result | main |    PR   | difference |
|    Passed   | 46897  | 46898 |     -1     |
|   Ignored   |  1634  | 1634  |     0      |
|   Failed    |  1723  | 1722  |     1      |
|   Panics    |   2    |   2   |     0      |

Fixed tests (1):
test/built-ins/Temporal/Duration/prototype/round/balances-up-to-weeks.js (previously Failed)

[jedel1043]

Oooh, this is actually really useful! IMO we should start doing something like this for all our spec-related methods.

(with enough tooling put into it, we could have something that keeps track of the oldest parts of the codebase that could be out of date with the latest specification, but that's just an idea)

[HalidOdat]

(with enough tooling put into it, we could have something that keeps track of the oldest parts of the codebase that could be out of date with the latest specification, but that's just an idea)

I like the idea of a tool for checking spec docs, maybe we should have something more structured so it's easier to parse and harder to mistype, maybe something like:

Spec(2025-05-16): <https://github.com/tc39/proposal-temporal/tree/c150e7135c56afc9114032e93b53ac49f980d254>

Or maybe just a date so it's easier to type, should be enough info to pin point a commit?

git log --oneline --since="2025-01-05" --until="2025-01-06"

[nekevss]

This looks good to me overall. Were you planning on looking into the abstract methods of round as well or were they all basically up to date?

[HalidOdat]

[..] Were you planning on looking into the abstract methods of round as well or were they all basically up to date?

I'll look into the abstract methods in a separate PR, don't want this PR to become too big 😅 From what I saw they where up-to-date at least the ones that where called directly in round.

[nekevss]

Sounds good to me. I'm fine with this merged as is. I'm mostly concerned about BubbleRelativeDuration. I think its bugged based off some of the tests, but it could be out of date as well.