I'm going to try to do a modest simulation of the Moon's libration as seen from the Earth (geocenter), similar to the technical one's I've shown in this question. I'll do it in Python and render it in Blender.
I'm familliar with getting Ephemeris type data from JPL Horizons and from Skyfield, both of which use the JPL Development Ephemerides (e.g. DE405,...), (see also Horizon's Ephemerides page and links therein). From those I generate local tables and interpolate them for animations and other non-critical calculations.
I know that at least some of the DE's contain lunar body motion (e.g. in DE405, but in DE406 it comes separately? Also discussed here) as Loading Auxiliary Files/Dynamic Frames, but without first-hand knowledge of reading the Spice Kernel's directly, is there a way I can get something close through Horizons? For example, with the position of a fixed position on the Moon with respect to Earth's Geocenter, or vice-versa in addition to the motion of the two centers I could reconstruct it well enough for a fairly accurate animation. However, this is as far as I've gotten - I don't know how to specify a lat/lon position on the Moon.
I could fudge it by just spinning the moon on its axis at a fixed rotational angular velocity, but if it's possible to get a handle on the actual lunar rigid body motion, that would be much more helpful to me in the long run.