add splitter config and SplitterOpsFactory to ExtractPDFFlow

[jojortz]

• Add splitter field to ExtractPDFConfig
• Create SplitterOpsFactory, which has two splitters:
  • ParagraphSplitter
  • MarkdownHeaderSplitter
• Create notebook to show pdf extract without split, extract_pdf_no_split.ipynb
• Test new config on the following notebooks:
  • extract_pdf.ipynb
  • pipeline_pdf.ipynb
  • extract_pdf_no_split.ipynb
• Update the following notebooks to have necessary instruction in GuidedPrompt:
  • openai_pdf_source_10k_QA.ipynb

[goldmermaid]

Nit: None is a better default value.

[goldmermaid]

I would suggest to always assume there will be a splitter for pdf flow for now. It is good to have this if-else case, but it can introduce extra complexity and in the long term cause inconsistency from with and without splitter pdf extractor. in this case, you should use a default splitter with for example paragraph splitter and allow people to configure it to other splitter such as markdown.

In the future, if there is a request regarding without splitter, we can think about bring it back. I would like to keep the code as simple as possible for now.

[goldmermaid]

In this case, you can directly use node = self._split_op(nodes) here.

[CambioML]

LGTM