My take on this is that the next MIME RFC should simply say that
the effect of escape sequences (or any format effectors other
than a CR LF pair) are undefined for the charsets it defines.
For charsets that require bi-directionality, define them in
a separate document.
Keith