Some tz.guess() tests are failing in 2024

[gilmoreorless]

Some tz.guess() tests are failing in 2024

[gilmoreorless]

The tests changed because of the following sequence:

1. When Intl is not available, tz.guess() falls back to checking a range of dates from <current year> - 2 to <current year> + 2. It looks for all zones that match the user's UTC offsets for that date, then chooses the zone with the largest listed population (to make the sorting stable).
2. Fiji stopped having DST after 2021. This included it in the <current year> - 2 calculation until the end of 2023.
3. Once the year ticked over to 2024, Fiji suddenly acted the same as a bunch of other Pacific zones without DST for the guess() function.
4. Since Fiji has a larger population than the other zones, it became the new default guess value, which failed the existing tests.

Given the low frequency of major DST changes like this, and given it's only for the non-Intl fallback, I don't think it's worth trying to change this long term. I've updated the tests in 49fa3a9, and we'll just have to keep an eye on the guess tests at the start of each year.