Multiple nested encodings is probably a good idea, at least to specify.
And the Content-encoding: list needs to be comma-separated if you're
ever going to be able to define anything that takes an argument, e.g.:
x-mumble-signature 13827423747234 239324237487234
as a legitimate ``content-encoding''. Yes, this suggests an
``encoding'' that is the identity transformation if the bits that flow
through it produce mumble-signature value 13827423747234 when encoded
with signed with parameter 239324237487234, but it's a legitimate desire
that the standard should allow.