For purposes of discussion, suppose that you arbitrarily split sequence serialization from single-item serialization into non-XML formats because I believe they are actually qualitatively different problems. Referring only to the sequence serialization side of the problem here, I think the question is whether XML sequence serialization and parsing has to in fact be consumable by an XML parser. As I see it, you either end up specifying some arbitrary set of privileged xml sequence tags:

<?xml version="1.0" encoding="UTF-8"?>
<xml:sequence xmlns:xs="http://www.w3.org/2001/XMLSchema">
    <xml:item value="foo" type="xs:string"/>
    <xml:item value="5" type="xs:positiveInteger"/>
    <xml:item type="document"><bar><bat/></bar></xml:item>
    <xml:item type="comment">foo</xml:item>
</xml:sequence>

or you work with a direct serialization as described earlier, possibly with RDF encodings for type:

(<?xml version="1.0" encoding="UTF-8"?>,"foo",5^positiveInteger,<bar><bat/></bar>,<!-- foo -->)