PyPDFToDocument: make conversion customization easier for users

[anakin87]

Is your feature request related to a problem? Please describe.
This stemmed from deepset-ai/haystack-tutorials#362.

Currently, to customize the PDF conversion process, the user has to provide a custom Converter (adhering to PyPDFConverter protocol).

While this allows great flexibility, it requires considerable effort for users who wish to customize only one extraction parameter (for example, extraction_mode). PyPDF extraction parameters

Describe the solution you'd like
It would be nice to provide a easier way to do simple customizations.

• Initially, I thought of allowing to pass extraction_kwargs in __init__ and also include them in the PyPDFConverter protocol.
• @silvanocerza proposed another idea: create something like a CustomConverter implementation (adhering to PyPDFConverter protocol) and make it possible for users to use it in a simple way.
  Something like:

  from haystack.components.converters.pypdf import PyPDFToDocument, CustomConverter
  
  custom_converter = CustomConverter(extraction_mode="layout")
  
  pypdf_to_document = PyPDFToDocument(converter=custom_converter)
  
  pypdf_to_document.run(...)

We would like to get @shadeMe's opinion on this...

[anakin87]

Passing converter allows almost unlimited customization, but I think this is rarely used and I argue that it would be easier for the user to create a custom component.

After an internal discussion, we decided to do the following:

• Deprecate the ability of passing a converter object at init time (and tell users to create a custom component if they need very specific customization). Remove this feature in future.
• Expand the customizability of PyPDFToDocument. We will follow the approach of PDFMinerToDocument, exposing a selected list of PyPDF parameters, chosen and serialized by us.

[anakin87]

done in #8569 and #8574