Inline IgnoreCase modifier does not work with non-ASCII

[stulov]

Hello!

I've noticed that inline i modifier clause does not transpile properly (at least as for my understanding) when dealing with non-ASCII characters. Here are some tests that fail.

{
	'pattern': '(?i:k)',
	'flags': 'u',
	'options': { modifiers: 'transform' },
	'expected': '(?:[Kk\\u212A])',
}

Here, /(?i:k)/u is expected to be equivalent to /k/iu, but while the latter matches '\u212A', 'K', and 'k', the actual produced expression /(?:[Kk])/u won't match '\u212A'.

{
	'pattern': '(?i:є)',
	'expected': '(?i:[Єє])',
}

Similarly, for 'є', to which only 'Є' case folds, the resulting expression /(?:\u0454)/ (0454 is a code point for 'є' that is, the expression is /(?:є)/) fails to match 'Є'.

{
	'pattern': '(?i:[є-ґ])',
	'expected': '(?:[Є-Ґє-ґ])',
}

Likewise, here the result is /(?:[\u0454-\u0491])/, which is still case-sensitive and basically just replaced symbols with the escape sequences for them.

I've tried to figure it out myself, but didn't manage to. I believe the problem is either somewhere around the caseFold and configNeedCaseFold* functions, or in the way iuMappings are generated.

Thank you!