The introduction paragraph of section 4.1.2 explicitly links to two
other RFCs for the missing syntax definitions,
RFCs 5234 and 5322 specifically.
and indeed the second one
(RFC 5222) contains a proper definition of atext.
Yep. The only thing that's unusual here is that these sorts of linkages are
usually declared in the conventions section of the document rather than waiting
until the "collected ABNF" section to say it.
I did not read the
whole of RFC 5222 recently, but a simple search found this definition
quickly and it seems pretty well-defined. Copying it over to RFC 5321
might be more convenient though.
Maybe, but these sorts of changes may not interact well with the ongoing EAI
work. I'm therefore opposed to making this change - if a change needs to be
made it would be to make the document linkage more prominent.