1. I concur with Neil and Don that we should define a binary subtype with
compression. (I don't expect it to get much use, but I think we long-ago
crossed over the line to try to provide the Internet with the tools for
doing 8-bit email transfer.)
2. I strongly suggest that the name of the two new subtypes NOT use
the string "compressed" or any other generic term. I believe it should
be a string which cites the SPECIFIC algorithm used for compression.
This makes it easier to add another algorithm, even though we hope not