Plugins will certainly be a useful addition. As far as the language goes, I think that while Python may be unwieldy, it's the only scripting system with the flexibility to make it worthwhile.
I only use Book View for previewing, but spent quite a lot of time fixing some of its defects and missing features for this release, as it's clear there are many who use it extensively. I think a move to TinyMCE would address a lot of the remaining issues, but will involve a fair amount of work.
Removing Xerces will actually speed things up quite a lot, but it's so deeply ingrained into the core of Sigil that it'll be an extensive change. I think Xerces offers advantages for handling pure xml components like the opf and ncx files, but the xhtml might be a candidate for a more simple string store.
As far as the professional users you mentioned in your other blog post, I think the fact is that when it comes to ebook production many of them are vague, full stop. The publishing industry as a whole has a rather low level of technical acumen when it comes to ebooks and this process is often farmed out to external conversion services who work to a price, which is reflected in the low technical quality of the majority of commercial ePubs. A large chunk of the minority who do know what they're doing have been working with
DocBook, which has been around for a long time, and they basically need XML schema translations, which is the main rationale for using OxygenXML to output ePubs (which is otherwise far inferior to Sigil as an ePub editor).