[Bee] Wrong structure and reading order in the form PDF file

### Bug
The results in the uploaded PDF document do not keep the structure of the doc, including its tables and content. All the contents are not in the proper reading order.

- sample file:
[002_USTaxForm_052_001.pdf](https://github.com/user-attachments/files/27866707/002_USTaxForm_052_001.pdf)

- sample md result:
[test_result.md](https://github.com/user-attachments/files/27866718/test_result.md)

### Steps to reproduce
process the sample file with docling DoclingParseV4DocumentBackend with `table model enabled`, and export different kinds of outputs.

### Docling version
docling==2.84.0
docling-parse==5.9.0
...

### Python version
3.12
...