Add ability to select page regions to be OCR'ed

If the PDF has two book pages per scanned page it comes out a mess; if there are images it tries to turn them into text. Unless you have nearly perfect clean scanned single pages expect to do a lot of cleanup of the results.