Extract all text from this document image.

For each text block, provide:
- "content": The exact text
- "type": header, title, paragraph, label, value, footer, or other
- "row": Group text on same horizontal line with same row number

Return JSON:
{
  "layout_rows": [
    {
      "row": 1,
      "elements": [
        {"content": "text here", "type": "header"}
      ]
    }
  ]
}