Add tests and/or fix large-magnitude numbers being converted to PluralOperands

[sffc]

From @markusicu:

How do you handle conversions from i128 and u128 types when you only have a u64 for the i, v, f operands? We truncate the high digits, and there are no CLDR rules for which that truncation makes a difference. But we can double check that this is indeed the case.

I should note that this also happens in string and FixedDecimal conversions. We should make sure we have consistent behavior across the board.

CC @zbraniecki

[sffc]

CC @younies

[Harsh1s]

Hi, I'd love working on this issue. Can you please assign this to me?