The second is that while a more simple XML markup, like what you loosely described as PGXML, sounds wonderful on the surface ... it requires, once again, largely reinvented the wheel
"Reinventing the wheel" is often something to be avoided, but I'm not sure it's a compelling issue here. First, there are other models to use, e.g. XHTML. Second, the most important standard is XML itself. That's what enables an incredible variety of tools and platforms; the specific DTD is much less important. (In fact, XML's designers made sure it was useful even without a DTD.) Third, TEI was created for a very different world: scholarly publishing. If PG's markup was going to be done by paid experts, TEI would probably be the best choice. But I'm not convinced it's appropriate for a volunteer organization. XML can be much simpler than HTML, yet TEI is (IMHO) more complex not less. I just finished converting The Wonderful World of Oz to PGTEI. (I'll post it on Classicosm.com once I have a chance to write up my impressions.) During my learning process, I came across an interesting comparison of Shakespeare marked up using TEI and an "ad hoc" markup used by Jon Bosak (a key inventor of XML). Though the comparison was done by a TEI advocate, I think Jon's is a much better model for our purpose. http://www.tei-c.org.uk/Sample_Manuals/mueller-main.htm A very gentle introduction to the TEI (the comparison is near the end -- look for the garish background colors)
Granted, the very nature of XML makes converting from a home-grown markup to TEI a possibility, removing the need to convert would seem to be the wiser path.
The whole point of a master format is that PG is going to convert to other useful formats. If TEI is useful in and of itself, that can be just another conversion. -- Scott Practical Software Innovation (tm), http://ProductArchitect.com/