king/cappallo may 1976   subroutine morpar
calculate partials of various constants in lunar librraion
equations with respect to parameters for which partials are
integrated

In principle, these derivatives should be set up also in parallel for
the orbit integration, but there would be no point in integrating the
orbit partial if the rotation partial were not done simultaneously.
Therefore, a single setup suffices, and the orbit uses pointers to
the derivatives organized by rotation order.

The only exception is the occasionally desired artificial case with
integration of orbit without rotation.  In this case, ROTINT is
false, and the rotation pointers would all be null, but it is
sufficient to fake the setup for the lunar J2 partial if the orbit
partial w.r.t. J2 is desired.