Apply Qwen NFC normalization before token splitting

Cargo.toml

@@ -35,6 +35,7 @@ sha2 = "0.11.0"
 shlex = "2.0.1"
 time = { version = "0.3.54", features = ["formatting", "parsing"] }
 turbovault-parser = "1.6.0"
+unicode-normalization-alignments = "0.1.12"
 ureq = { version = "3.3.0", default-features = false, features = ["rustls"] }
 url = "2.5.8"
 usvg = "0.45.1"