May be I just didn't read RFC 821 and RFC 822 carefully enough, but
I couldn't find anything about whether the 1000 characters line limit
should be applied to RFC 822 headers before or after unfolding.
The "1000 character limit" (actually, the lowest value at which it is
acceptable to set a maximum) is a transport limitation in RFC821.
Because it is a transport limitation, it would apply to the folded form.
To the best of my knowledge and recollection, you will find no
restrictions at all on either the total number of characters in an
RFC822 field (e.g., "unwrapped To") or, isomorphically, on the number of
RCPT TO fields that can appear in the envelope in RFC822 or 1123, which
are the only official/standard places such a limitation could appear in
the present state of the protocols (other than 1123, all other
"extensions" and "clarifications" to 822 so far are of the nature of
opinions, not standards).
I think it would be consistent with common practice and good
judgment for a UA to start discarding comments and, if necessary,
personal names (i.e., "phrase" in addresses) if they became long and/or
obnoxious enough to start causing problems. But, ultimately, to impose
a limit on the "unfolded" "To:" line implies imposing a limit on the
number of addressees for a message, and the RFCs do not anticipate such
a limit.
Unless there is an extension to be proposed here to Internet mail, I'd
suggest that further discussion might fruitfully be moved to
header-people(_at_)mc(_dot_)lcs(_dot_)mit(_dot_)edu (subscription requests to
header-people-request, as usual, and be patient--the maintainers seem to
be sufficiently overextended to not have time to follow up on accepted
offers of help). The readership of that list seems to thrive on
questions like this, does not restrict itself to Internet issues, and
could probably provide you with advice about what others have done under
similar stresses.
--john