[Feature Request]: Split DeepDoc to an independent program

[firezym]

I have tested and compared the DeepDoc module with [marker-pdf](https://github.com/VikParuchuri/marker), which has 20k stars and focuses solely on parsing PDFs. I think your OCR quality is better, especially for charts and tables. You could consider splitting the module out as a standalone program and offer chargeable online saas service for parsing PDFs into Markdown, similar to how [Jina Reader](https://jina.ai/reader/) converts HTML to Markdown.

A standalone service could also contribute to the RAGFlow agent modules, as users might sometimes need to transform PDFs into Markdown for further use with LLMs without necessarily storing them in the knowledge base.

[Snify89]

It is already somewhat splitted.
The only issue is, (western) European characters are not applied (Umlaute, etc.)