syntax: fix negation handling in HIR translation

[pandaman64]

Previously, the HIR translation of ClassUnicode ignored the != operator in the ClassUnicodeKind::NamedValue variant as it referred to ClassUnicode::negated, which describes whether the class uses \p or \P.

This commit fixes the issue by calling ClassUnicode::is_negated.

Bug reproducer

Playground

use regex_syntax;

fn ast_parse(pattern: &str) -> regex_syntax::ast::Ast {
    let mut parser = regex_syntax::ast::parse::Parser::new();
    parser.parse(pattern).unwrap()
}

fn main() {
    // AST parser correctly parses the != operator
    println!("{:#?}", ast_parse("\\p{gc=separator}"));
    println!("{:#?}", ast_parse("\\p{gc!=separator}"));
    
    // The HIR translation forgets !=
    let hir1 = regex_syntax::parse("\\p{gc=separator}");
    let hir2 = regex_syntax::parse("\\p{gc!=separator}");
    println!("hir1 == hir2: {}", hir1 == hir2);
    println!("{:#?}", hir1);
    println!("{:#?}", hir2);
}

• Expected behavior: hir1 == hir2 should be false.
• Actual behavior: hir1 == hir2 is true.