[refurb] Handle non-finite decimals in verbose-decimal-constructor (FURB157)

[dylwil3]

This PR extends the Decimal parsing used in verbose-decimal-constructor (FURB157) to better handle non-finite Decimal objects, avoiding some false negatives.

Closes #14587

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.

[MichaReiser]

Unrelated to your fix: To avoid an allocation in many cases, it might be worth trying an exact match first before calling to_lowercase.

let trimmed = float.value.to_str().trim();

if is_inf_infinity_or_nan(trimmed) {
	return;
}

let normalized_float_string = trimmed.to_lowercase();

if is_inf_infinity_or_nan(&normalized_flaot_string) {
	return;
}

[dylwil3]

Maybe I'm misunderstanding but I think we have to allocate anyway: we only get to escape early here if float.value is not inf, infinity, or nan (after trimming and lowercasing). So if we first checked that the trimmed version wasn't "nan" etc., we would still have to rule out the case where it's "NaN".

[MichaReiser]

Yes, it's just that we do it less often. I guess the overall idea is to spell out all common variations so that the rule can short-circuit for almost all valid-cases and only allocates for uncommon spellings (e.g. nAn).