LibJS: Handle Unicode ID_Start characters excluded from XID_Start

[officialasishkumar]

The ECMAScript specification defines UnicodeIDStart as any Unicode code point
with the ID_Start derived property. The unicode_ident crate that the Rust
lexer uses implements XID_Start, which is a proper subset of ID_Start.

The two sets differ for characters whose NFKC normalization form does not begin
with an ID_Start character. These characters are valid JavaScript identifier
starts per spec but were rejected by the lexer with an Invalid token.

The existing code already worked around this for U+309B and U+309C (Katakana
voiced/semi-voiced sound marks). This commit extends that workaround to cover
all 21 code points in ID_Start \ XID_Start under Unicode 16:

Code point	Name
U+037A	GREEK YPOGEGRAMMENI
U+0E33	THAI CHARACTER SARA AM
U+0EB3	LAO VOWEL SIGN AM
U+309B	KATAKANA-HIRAGANA VOICED SOUND MARK (already handled)
U+309C	KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK (already handled)
U+FC5E–U+FC63	Arabic shadda ligature isolated forms
U+FE70, U+FE72, U+FE74, U+FE76, U+FE78, U+FE7A, U+FE7C, U+FE7E	Arabic vowel isolated forms
U+FF9E	HALFWIDTH KATAKANA VOICED SOUND MARK
U+FF9F	HALFWIDTH KATAKANA SEMI-VOICED SOUND MARK

The unicode_id_continue function is similarly updated for those characters
whose NFKC form also contains a non-ID_Continue character (a space), so they
remain valid as identifier-continuation characters. U+0E33, U+0EB3, U+FF9E,
and U+FF9F are omitted from that list because their NFKC forms consist solely
of XID_Continue characters and are therefore already handled by the crate.

Fixes #8870.

[ladybird-bot]

Hello!

One or more of the commit messages in this PR do not match the Ladybird code submission policy, please check the lint_commits CI job for more details on which commits were flagged and why.
Please do not close this PR and open another, instead modify your commit message(s) with git commit --amend and force push those changes to update this PR.

[trflynn89]

oof, a couple things here (see #8788 (comment) for some info)

1. Let's not use unicode_ident at all. As pointed out in the link above, we should avoid duplicating large Unicode tables.
2. That PR added FFI to just invoke Unicode::code_point_has_identifier_start_property which should already be doing the right thing. Let's do the same here. That should let us remove the unicode_ident package from Cargo.toml.
3. The comments added here are exceptionally verbose.

[officialasishkumar]

Will update accordingly.