I don't believe the ordering requirement is reasonable. Sounds like a
Pine bug, plain and simple.
By the way, the PrintHeader() routine in the current metamail release
has a bug in the handling of white space. That bug will be corrected
in the next release, coming (I hope) next week.
Finally, I believe Pekka said "RFC 1344" where he meant "RFC 1342",
in case anyone was confused.
I'm glad to hear about MIXMH. I'm not sure what "limited support for
multipart" means, but have you considered having a fallback strategy
where messages of unrecognized types get passed off to metamail?
That's what I do in the MIME version of Andrew -- I built in support
for many MIME types, and I put in click-buttons to call metamail
for any unrecognized types, since I assume there will always be some...