I had an identical issue yesterday using the same mission architecture, and have spent most of today working on tracking this down. I don't yet have a solution or workaround, but I can at least narrow it down somewhat. Here's what I've got:
Like you, I built an apollo-style Mun mission, with a LES on top, then Apollo-style CSM assembly, then Munar-lander with separate ascent/decent stages, and then three stages underneath that. Several things bugged out and im not sure which of them are separate bugs, and which of them are the same bug manifesting multiple symptoms. In order of appearance:
Fuel drained from the decent stage of my lander when the third-stage engine of my burned, even though they were separated by a decouple.
After CSM separation for the Lem+Transfer stage assembly, while controlling the CSM, right clicking on the Lem's docking port did not give me an option to set it as target. Quicksaving and quickloading at that point Does restore the option to target, but Does Not Prevent any and of the other bugs, which appear regardless of whether I quicksave/quickload or dock without targeting and continue the mission.
Roll and Pitch control authority on the reaction wheels is lost sometime between the munar transfer burn and arriving in the muns SOI. Yaw continues to work, and RCS and Engine gimbaling still have full freedom of movement on all axes.
After munar capture, undocking the craft while not under timewarp results in a vessel destroyed message. Despite this, going to the tracking station shows that the vessel is still there and can be controlled, however the docking ports no longer function at all.
If instead, I try undocking *while* under time warp, one or both new craft get a prograde acceleration of about 500 m/s.
My first thought was that this was a rare bug involving all docking ports but only affecting me and a small handful of others, so to test this I spent most of this morning building a small orbital station. The docking ports were a little... imprecise? sometimes registering connection when the craft were at very wide angles, or very far apart, but docking and undocking worked without destroying anything, I can therefore rule out a general docking port bug.
Im currently testing the lander legs. The feul flow issues with those are pretty bizarre, and seem like an obvious place to test next; Ive read elsewhere in the forum that putting the legs on radial decouples prevents the feul flow issues, so I did that. If it works, I'll see if I still bug out when trying to undock.
If I do, I'll build a lander that lands on struts - no legs at all - and see what effect that has.
If the bug persists I'll build a CSM version with radial engines so I don't need to flip it around for its transfer burn, and see if its the flipping or the redocking to the same craft you launched on that causes it.
Depending how that goes, I'll probably have more station testing to do.