feat: Add get_image for all DocItem, specialize for FloatingItem

[sh-gupta]

We want the image attribute to be available for TextItem to allow exposing equation images. I have moved image from FloatingItem to DocItem so that it becomes part of TextItem as well.

Related issue: DS4SD/docling#299

[vagenas]

@sh-gupta @cau-git ⚠️ note that any changes to the subtypes of DoclingDocument:

1. Need to be backwards compatible within docling-core 2.x.y, e.g. new field with default value, and
2. then imply a new minor version bump of the DoclingDocument as well as potentially DocMeta which also depends on the DocItem.

[PeterStaar-IBM]

@sh-gupta We quickly discussed, we need to do the following points (@cau-git will explain),

• add get_image to DocItem (overloaded on PictureItem)
  • check if there are page-images
• enrichment models can leverage the backend
• deprecate this: https://github.com/DS4SD/docling/blob/main/docling/datamodel/pipeline_options.py#L84C5-L84C28

[sh-gupta]

Made the required changes:

1. Moved image from DocItem back to FloatingItem
2. Added get_image function to DocItem. If returns None if provenance is not found or page image is not available, else it returns an appropriately cropped section of the page image
3. Overrode get_image in FloatingItem

[cau-git]

Made the required changes:

1. Moved image from DocItem back to FloatingItem
2. Added get_image function to DocItem. If returns None if provenance is not found or page image is not available, else it returns an appropriately cropped section of the page image
3. Overrode get_image in FloatingItem

This looks good now, thanks! To move on, you best create a companion PR on main docling where you pin this branch of docling-core to test it further.

[PeterStaar-IBM]

@sh-gupta Could we add some unit-tests for this functionality?

[dolfim-ibm]

Looks good.
I'm wondering if we want to raise an exception instead of returning None.

[PeterStaar-IBM]

very nice!

[sh-gupta]

@PeterStaar-IBM : I've added tests for DocItem.get_image and FloatingItem.get_image

@dolfim-ibm : I initially thought raising an error but then decided to stick with None to be consistent with the behavior of FloatingItem.image when image is not found. I can raise an exception if you prefer.

@vagenas : I've made the changes as per your discussion. Can you please take a look?