Parallelize DFS, use an AtomicBool for short-circuiting

time:   [73.407 µs 76.145 µs 79.345 µs]
change: [−40.126% −24.679% −3.8062%] (p = 0.04 < 0.05)

src-tauri/Cargo.toml

@@ -30,3 +30,19 @@ log = "0.4.29"
 rand = "0.9.2"
 docx-rs = "0.4.18"
 anyhow = "1.0.100"
+rayon = "1.11"
+
+[dev-dependencies]
+criterion = { version = "0.8.1", features = ["html_reports"] }
+
+[[bench]]
+name = "rayon"
+harness = false
+
+[lints.clippy]
+indexing_slicing = "deny"
+fallible_impl_from = "deny"
+wildcard_enum_match_arm = "deny"
+unneeded_field_pattern = "deny"
+fn_params_excessive_bools = "deny"
+# must_use_candidate = "deny"