Big thanks to the DEVONthink team. With the last version (3.8.1) the problem has been solved, that PDF/A documents in particular contained no or hydroglyphs as text after editing (highlighting, …, or combining).
Great!
BTW: Next step could be to show, if a document is PDF/A (as in Adobe Acrobat, see screenshot)…
Logic suggests that, since the problem was also in Preview, the macOS 12.2 update fixed it, not the DT 13.8.1 update.
I had not noticed this, but the posting here is a good catch.
I agree that showing PDF/A (versus just PDF) in DT would be beneficial. The former is supposedly the preferred standard for archival documents (hence the /A). It helps also to see that PDF/A compliance requires that you include such things as a table of contents in the document, as well as the meta-content for author, title, and so forth. Great stuff to assemble during LaTeX file generation (as a side bar).
Finally, note that various standards exist for /A as shown here PDF/A - Wikipedia
Perhaps DT could have access to show this information and maybe even provide a way to have it used in a search criteria (e.g. type = pdfa-3 versus pdfa-2)
Is all that really a part of the general PDF/A specification or rather part of the conformity level B? I think, a PDF/A can be perfectly valid even without a TOC and metadata (think a PDF/A invoice, where neither of those are very useful, in my opinion)
Yes, correct. I did follow up reading. PDF/A at various levels only assures future-proofing to access the content. It is not a standard on the content or its organization.