Fix TypeError when combining the text table strategy with text_layout

[MohammadHijjawi97]

Problem

The table-extraction docs say that all text_-prefixed settings are passed to text extraction, and that "All possible arguments to Page.extract_text(...) are also valid here." With the "text" strategy, though, some of them crash:

page.extract_tables({
    "vertical_strategy": "text",
    "horizontal_strategy": "text",
    "text_layout": True,
})
# TypeError: WordExtractor.__init__() got an unexpected keyword argument 'layout'

The same happens with other extract_text-only arguments (e.g. text_x_density, text_layout_width). With the default "lines" strategy, the same settings work fine.

Cause

For the "text" strategy, TableFinder.get_edges() calls self.page.extract_words(**settings.text_settings). extract_words passes everything to WordExtractor, which only accepts word-extraction arguments.

Fix

When finding words for the text strategy, only pass on the text_settings keys that WordExtractor accepts. These are listed in utils.text.WORD_EXTRACTOR_KWARGS, which chars_to_textmap and extract_text already use for the same purpose. Cell text extraction still receives all text_* settings, so text_layout etc. keep working there.

Tests

• New test_text_layout_with_text_strategy in tests/test_table.py, using the existing senate-expenditures.pdf. It fails on develop with the TypeError above and passes with this change. It also checks that the layout-extracted cells, once stripped, match the non-layout ones.
• tests/test_table.py passes.
• black --check, isort --check-only, flake8 and mypy --strict --implicit-reexport all pass.

CHANGELOG updated.