fix: ocr when no elements are found in block

[qued]

Stopgap fix for a bug that causes the parsing procedure to ignore pdf elements that are not contained within the bounds of an inferred/specified layout element.

Testing:

With the following linked file, loremipsum-flat.pdf, try the code:

from unstructured_inference.inference.layout import DocumentLayout
doc = DocumentLayout.from_file('loremipsum-flat.pdf')
print(' '.join([el.text for el in doc.pages[0].elements]))

Observe that the output is blank on the main branch, but contains text on this branch.

[benjats07]

LGTM