[enhancement] Return the detected widgets in the `prepare_form` function

[ValeriL]

Hi there,
For some tasks, it is necessary to know which widgets were detected and where they are located.
In my opinion, it would be convenient if the function additionally returned the results obtained during the initial pass, in order to avoid re-parsing the PDF using alternative libraries.
I mean results from this row
https://github.com/jbarrow/commonforms/blob/main/commonforms/inference.py#L267