barring any comment from Junio, I think a fixup is probably more
onerous than a resend with the fixup squashed in; that is, assuming
the old iteration is already git-am'ed in, and the fixup applies onto
a new iteration.
I'll clarify. The fixup applies to `d8e255cb385365` in `pu`. The
thread begins with a patch generated from this commit: the commit is
stalled in `pu` due to lack of acks.
-- Ram