There's hope for your wife! My elderly mother had a side job pre-retirement setting up computer systems for small businesses (like, server plus maybe a dozen workstations), and through me got into Linux. Now she rolls Linux Mint, and got my dad -- who first thought my installation of OpenOffice on his Windows machine was "Linux" -- onto Ubuntu. Now neither one of them will ever look back.
To address the argument at hand: how one negotiates with commercial publishers on submitting jobs is outside the scope of the questions surrounding how Lily might be improved to export to whatever format they insist on, and the resolution of that issue depends more on matters of ethics and philosophy than technical matters of toolchain management. I've done plenty of work, for example, for commercial (text) publishers who have requested submissions in .doc format. I've explained to them that I do not have native Word, but can export from LibreOffice (with the caveat that I can't guarantee the formatting will match what I see locally), and I've never had an issue. I honestly think it's inappropriate for a contract-giver to dictate to me what tools I use locally, so long as my output can match their standards to a reasonable degree (and especially since they do their own formatting in-house regardless).
The problem here is that neither Sibelius nor Finale is ever going to have development work done on making sure its MusicXML *import* will work properly: iirc (Urs, correct me if I'm wrong here), Sibelius' publisher fired all its core developers in 2012, so they have no one with the technical expertise with the core code to make such changes. Whether Dorico (which is where those devs went) implements such functionality is an open question (though, I'd be pretty certain that the terms of their departure include a non-compete clause, so interoperability with Sib would *still* be foreclosed).
But I think it is much more in *our* interest to focus on getting export from .ly files into any number of other formats (but chiefly MusicXML) to work better, so that these issues don't come up. This benefits not only Lily engravers who have to work with commercial publishers locked into Finale/Sibelius, but also would greatly aid our future work with, for example, the MEI. After my experience this week -- in which we found that producing a PDF with Lily, then using an OCR (SharpEye, with many thanks to Phil Holmes who did the work) to output to Sib worked much faster and better than Frescobaldi's MusicXML export (though this may be my own fault, as the .ly files are rather convoluted) -- I would really love to see that side of Lily's workflow improve.
Cheers,
A