2021-09-22:
Paul Levy pointed out that the definition of $\psi^{-},\psi^{+}$ on
page 88, column 2 is incorrect, because the righthand sides of the
definition is not necessarily admissible (because $R^{-},R^{+}$ are
not assumed to be admissible).
I believe the fix is to take the admissible hull of the existing
righthand sides as the actual definition of $\psi^{-},\psi^{+}$. This
now does give a monotone operation on $\mathcal{P}$, whose least fixed
point $\phi^{_},\phi^{+}$ still has the crucial property that the set
of $(e^{-},e^{+})$ defined at lines -5 and -4 of p88,col2 is closed
under the action of $\delta_{D^{-},D^{+}}$ -- so the proof before the
statement of Theorem 7.3 still goes through for this new defintion of
$\psi^{-},\psi^{+}$.