Norm describes the issue of "caption" in HTML tables and his move to
make caption mixed content to support it.
Other than adding the HTML table tags, the DocBook DTD doesn't
include the HTML tags, so we are not talking about mixing HTML
elements in here (unless you make some local customizations
to do so, of course).
I never really thought that any HTML elements other than table elements
were added; I just somehow got confused about "mixed content" as in XML
(elements plus text [xml]), mixing inlines and blocks, mixing HTML table
elements with DocBook elements, etc.