Support for alternative PDF parser

[cilia]

Is there plan to support alternative PDF parsers such as PyMuPDF? Maybe as an option to let user specify and try different parsers so they can decide which is the best to use for their specific PDFs?

[dhdaines]

This is difficult to do because Camelot depends on pdfminer.six's layout analysis.

That layout analysis algorithm could most likely be implemented on top of pypdfium2, pikepdf, pymupdf, etc, but nobody has done this yet. It will be difficult to have the exact same results because there are some idiosyncracies (some might say deficiencies) in the way font metrics are handled by pdfminer.six.

There is an alternative implentation on top of PLAYA-PDF though, which gives a 10-25% speedup, see #589