> thus implying e.g. an optional 'data' value. This is obviously no CSS
> matter, and so there is no need for use of the class attribute, and the proposed

class is not "CSS class", or even more generally, style class.  That sort
of thinking leads to abuses like class="red".  class is a semantic sub-classing
of an element, which refines its meaning without changing the basic meaning.

As such, class is the best available attribute for indicating layout tables,
except that layout tables are really styling and the intention was, as I 
understand it, that by now would be obsolete because styling languages would
have taken over the function.

Defining a new attribute (with a five to ten year lead time for
implementation) as a workaround does not make sense to me, as a way of
working round the apparent ten year lead time for implementing separation
of styling properly.

(The normal use of type is for MIME media types.)