AmazonTextractPDFLoader should allow us to configure the TextLinearizationConfig

[lvieirajr]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

Currently the AmazonTextractPDFLoader uses a default set of configurations for how to linearize the output from textract.

It instantiates the TextLinearizationConfig internally without the ability to receive parameters from the caller:

linearizer_config = self.textractor.TextLinearizationConfig(
            hide_figure_layout=True,
            title_prefix="# ",
            section_header_prefix="## ",
            list_element_prefix="*",
)

It would be great if when instantiating the PDFLoader we could pass in a LinearizationConfig object so that we could add things like ignoring footers, headers, and any other of the many available configurations for the linearizer.

Motivation

The motivation behind this is because once the document is loaded I only get the strings for each page as output so I lose the option to customize anything on my end based on the textract output itself cause that is used and is not stored anywhere for reference.

Proposal (If applicable)

Add a new parameter to AmazonTextractPDFParser called linearization_config that is not required but if passed it will be used instead of the one that is instantiated inside of it.
Add the same parameter to AmazonTextractPDFLoader and pass it down to the AmazonTextractPDFParser when instantiating it.