The metamail program itself never generates any mail, so it is clearly
not a culprit.
Other programs included within the metamail distribution do generate mail,
and some of them have been culprits in the past. In the first release,
some of those programs actually generated dots in tspecials. In the second
release, I think they didn't, but there were some places where users would
supply a content-type and my code didn't check it & quote things where
necessary. I don't know of any such problems in the current release, but
there may well be some. (Indeed, the metamail distribution now includes
some contributed software that I can't vouch for, and frankly I'm reluctant
ever to vouch for my OWN code...:-)
Having said all that, I would say that rejecting mail with unquoted dots
is completely reasonable, and is the best way to make sure that such
errors get caught sooner rather than later... -- Nathaniel