I have done some more testing of schema-aware validation of XHTML
output and made the test into an article:

"Valid XHTML with schema-aware XSLT 2.0".
http://www.xmlplease.com/validxhtml

If I've read it correctly you seem to be complaining about defaulted
attributes (in both articles) - DTDs and XML Schemas have done that
for a long time and I dislike it too, but it's nothing to do with
schema aware XSLT, just a by-product of validation.  You can't ask
validators to "not copy that junk, please".  If you don't want those
values defaulted, you need to get the schema changed (or perhaps use
the Relax NG versions).

I do think though that schema aware transforms to XHTML is probably
where its biggest use  of validating the result will be.  It's early
days yet - once you've got your valid XHTML you still have to get IE
to display it properly ;)

I've always liked the idea of a schema aware processor that given an
input XML and a schema, could report XPaths and templates in my
stylesheet that would never select anything - say because of a typo,
or because of an update to the schema...