Why XML? In the context of the CHM format I suspect because the answer to everything around that time was XML. If XML wasn't solving your problem you needed more of it.
When CHM arrived on the scene, ZIP was far from being treated like container format, and definitely not ubiquitous.
As for XML - it's ch easier to make a stable parser that is complete and fast for XML than it is for SGML-based HTML, and indeed the was a big push towards XML everywhere... Partially because of that reason.
Because technical writing is about structured authoring (e.g. using specialized content items like step-by-step guides, FAQs or Quizzes) and I follow this metaphor, avoiding excessive formatting options and concentrating on things that matter.
Epub suffers from the same flaws - only one language per file, freeform styling that makes company-wide style consistency problematic (and when we talk about a ERP system with thousands of users and a miriad supplementary tools created for it - it's a really big problem).
And it was not designed for context-sensitive help of desktop apps, while LiteHelp is.
Everybody knows how to author html, if xml is some specific new set of rules, what is the advantage of that approach?
If it is xml, what makes it different from the epub format? Is epub format the subset? What were the advantages of it not being the subset?