PDFs don't have "junk PostScript", they have PDF object(s) which may or may not have been made w/ PostScript code, and possibly PostScript XObjects which are intentionally created and placed in the stream of other objects and which certain PDF renderers (usually on imagesetters) could make use of, unless the software doing the distilling to PDF is either technically wrong and not adhering to the PDF specification, or maliciously hiding PS code (though w/o a Display PostScript or other PS interpreter, I fail to see the usage in doing so).
How tables are constructed in a PDF (or designed for that matter) can vary pretty wildly, and apparently there is some added tagging involved in PDFs being sent to the KS via e-mail so that they can be marked up (perhaps a resurrection of the "Print Replica" format?).
Once I get mine, I'll see if such files can be transferred off, and start disassembling them (Fortunately the PDF spec is available:
https://opensource.adobe.com/dc-acro...ence1.7old.pdf and
https://web.archive.org/web/20220306..._17_errata.pdf --- though why those twain couldn't be integrated into a single updated document, I'll never understand).