The primary drawback

 to HTML is that it is inherently a multi-file solution; 

I'd say that's far from the primary drawback.

Much more substantial drawbacks are that is presentational,

not syntactic; and even if you make it even more complex

with syntactic information (or don't for that matter) the

proofers will never (nor should they) proof in that format.

For DP's purposes, for actually doing the work, HTML is

a non-starter - but so is any other equally complex (I'd say

any XML-based) representation. What we have in there

already (<i> etc.) is the locus of major headaches and

an ongoing error-trap.