Remove tests that rely on nonexistent bad-option error

[catamorphism]

I suggest we remove these two tests for now, because the property that they are testing is actually underspecified. There is nothing in errors.md corresponding to the bad-option error, and what's more, the spec doesn't require :number to signal an error for a non-integer minimumFractionDigits value.

I filed #738 and #739 to clarify those issues in the spec, but I think the tests should be removed in the meantime, and re-added once the behavior is fully specified.

[aphillips]

I think this would be a mistake. I agree that there is no specific bad-option error. I think we should fix registry.md for some of the cases you point out. I think that number formatting is so common that these options should be well-clarified now--and that there will be general agreement about how to do that. I think these tests should be retained.

[aphillips]

Note: I made #741 to (try to) address part of the gap. It doesn't mess with the bad-option problem.

[eemeli]

I don't think these should be dropped. The right fix here would be to better define the behaviour for :number minimumFractionDigits=foo, which should indeed be considered an error.

[aphillips]

I'm marking this for Monday's agenda. I used resolve-candidate because at least two of us feel that this shouldn't go in, but that doesn't mean the WG won't feel differently.

[aphillips]

In the 2024-03-25 call we agreed not to merge this.