include_converted_doc api chunking | Not Return Converted Document

### Bug
include_converted_doc api chunking | Not Return Converted Document
...

### Steps to reproduce
Run latest version of Docling Server Cpu

run as python or other like as follow:

```python
def upload_file_for_chunking_task_async(file_path: str) -> dict:
url = f"{DOCLING_URL}/chunk/hybrid/file/async"
logger.debug(f"[POST FILE] {url} – file: {file_path}")

try:
with open(file_path, "rb") as f:
files = {"files": (os.path.basename(file_path), f, "application/pdf")}
data = {
"include_converted_doc": "true",
"target_type": "inbody",
"convert_do_ocr": "true",
"convert_force_ocr": "false",
"convert_ocr_engine": "auto",
"convert_ocr_lang": "it",
"convert_image_export_mode": "placeholder",
"convert_pdf_backend": "dlparse_v4",
"convert_table_mode": "accurate",
"convert_table_cell_matching": "true",
"convert_pipeline": "standard",
"convert_do_table_structure": "true",
"convert_include_images": "false",
"convert_images_scale": "0.1",
"convert_md_page_break_placeholder": "",
"convert_do_code_enrichment": "false",
"convert_do_formula_enrichment": "false",
"convert_do_picture_classification": "false",
"convert_do_picture_description": "true",
"convert_picture_description_area_threshold": "0.05",
"chunking_use_markdown_tables": "true",
"chunking_max_tokens": PARSER_MODEL_MAX_TOKENS,
"chunking_tokenizer": PARSER_MODEL_TOKENIZER,
"chunking_merge_peers": "true",
}

resp = requests.post(url, files=files, data=data, timeout=600)
logger.debug(f"[RESPONSE STATUS] {resp.status_code}")
logger.debug(f"[RESPONSE BODY] {resp.text[:800]}")
resp.raise_for_status()
return resp.json()

except Exception as e:
_log_request_error(e, url)
raise

```
...

### Docling version
v1.8.0
...

### Python version
n/a
...

Start by reproducing the request against the /chunk/hybrid/file/async endpoint using include_converted_doc=true and the supplied conversion and chunking options. Inspect the endpoint's response handling to determine why the converted document is absent; done means the response includes the converted document for this request, with regression coverage added if the project provides a relevant test entry point.

Written by the indexing model from the issue text.