First of all, the use of ISO-2022-JP is specifically limited in its
2022 functionality so that is it does fulfill the requirements of being
a character set. This in no way implies that ISO-2022 is itself a character
set. It is not. The working group decided to punt on the issue of how to
specify ISO-2022 usage, since the need for it was not clear and the working
group did not feel competent to write such a specification.
What you've done here is propose an extension and then complain that the
resulting usage does not match the wording of the specification. You're
right in saying that broader use of 2022 as part of a charset parameter
would violate the definition of a parameter. However, the limited usage
we've specified does match the definition.
All this does not mean that there's no place for other uses of 2022 in
MIME. You should feel free to write a specification of how to specify 2022
usage in MIME. Just don't try to pack it all in the charset parameter -- that
is certain to be wrong. My personal favorite way is to have 2022 as a
subtype of text and then have parameters to specify the specifics of the
initial 2022 state.
Don't expect to see this material incorporated into the base MIME document.
However, just because it won't be in the base document is no reason to
refrain from doing it. The whole point of MIME is that it is extensible, and
we expect to see lots of extensions made to it.