chore: track *.traineddata via LFS + Gill Sans font->model mapping

.gitattributes

@@ -34,3 +34,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 smoke_signal/tessdata/gill_sans_infant.traineddata.otf filter=lfs diff=lfs merge=lfs -text
+*.traineddata filter=lfs diff=lfs merge=lfs -text
+*.otf filter=lfs diff=lfs merge=lfs -text

smoke_signal/scripts/font_library.py

@@ -275,6 +275,16 @@ def register_font_model(
     return {"ok": True, "font_name": font_name, "tesseract_model": model}
 
 
+# Static font name → model name mapping.
+# Handles cases where the slug doesn't match the traineddata filename.
+FONT_MODEL_MAP: dict[str, str] = {
+    "gill sans infant":  "gill_sans_infant",
+    "gill sans":         "gill_sans_infant",  # MixFont may return either name
+    "gill sans mt":      "gill_sans_infant",
+    "gillsans":          "gill_sans_infant",
+}
+
+
 def resolve_tesseract_lang(font_name: Optional[str]) -> str:
     """
     Return best tesseract language/model to use for this font.
@@ -283,6 +293,13 @@ def resolve_tesseract_lang(font_name: Optional[str]) -> str:
     if not font_name:
         return "eng"
 
+    # Check static map first
+    mapped = FONT_MODEL_MAP.get(font_name.lower().strip())
+    if mapped:
+        tess_dirs = [TESSDATA_DIR, REPO_TESSDATA_DIR]
+        if any((d / f"{mapped}.traineddata").exists() for d in tess_dirs):
+            return mapped
+
     registry = _load_registry()
     entry = registry.get(font_name, {})
     candidates = []