Match official demo message format: text before image

Update message format to match LightOn's official demo implementation:
- Add empty text element before image in content array
- Order: text first, then image (matching their Gradio demo)

This ensures we're using the exact same message structure as their
production demo, which should improve reliability and quality.

Reference: Official demo at https://huggingface.co/spaces/lightonai/LightOnOCR

🤖 Generated with Claude Code

Co-Authored-By: Claude <noreply@anthropic.com>

lighton-ocr.py

@@ -137,11 +137,13 @@ def make_ocr_message(
     pil_img.save(buf, format="PNG")
     data_uri = f"data:image/png;base64,{base64.b64encode(buf.getvalue()).decode()}"
 
-    # LightOnOCR uses simple message format (no explicit prompt text needed)
+    # LightOnOCR uses message format with empty text prompt before image
+    # (matching official demo: text first, then image)
     return [
         {
             "role": "user",
             "content": [
+                {"type": "text", "text": ""},
                 {"type": "image_url", "image_url": {"url": data_uri}},
             ],
         }