Description
Given that there are some problems with how overlapping text is
removed (slow performance: PDFBOX-956; some chars incorrectly skipped:
PDFBOX-1155), I think we should make this controllable from Tika's
PDFParser and I think it's best to default it to off for now.