Convert \u{} in identifiers into UTF-8

[strager]

When quick-lint-js encounters an identifier with an escape sequence, it replaces the escape sequence with the escaped character:

let a = "hello";
console.log(\u{61});  // Prints 'hello'. '\u{61}' is replaced with 'a' in-place during lexing.

Currently, we only support ASCII characters. We should support non-ASCII characters:

let \u{0220} = "hello";
console.log(\u{61});  // Prints 'hello'. '\u{0220}' should be replaced with bytes 0xC8 0xA0 in-place during lexing.

Note: Recognizing unescaped non-ASCII characters is out of scope for this task.

[strager]

Landed in commits c960e7a and 340c0da.