Bug Description

Negated Unicode property class \P{Han}+ matches Han characters instead of excluding them.

Reproduction

pattern := `\P{Han}+`
input := "abc中文def"

// stdlib: ["abc", "def"]  (correctly excludes Han characters)
// coregex: ["abc中文def"]  (incorrectly includes Han characters)

Expected Behavior

\P{Han} should match any character that is NOT in the Han Unicode script.
On input "abc中文def":

• "abc" matches (not Han)
• "中文" does NOT match (is Han)
• "def" matches (not Han)

Actual Behavior

The entire string matches including Han characters.

Root Cause

Likely in how negated Unicode property classes are compiled in nfa/compile.go.
Need to investigate compileUnicodeClass or related Unicode handling.