It isn't so much that the Target's Ap and Pe are oscillating - it's the way OrbitMFD works. OrbitMFD calculates the orbit based on the bodies current velocity vector - and the calculations are a bit simple so the "projection" doesn't predict the perturbations caused by non=spherical gravity.
The Target will be at very close to the same altitude, and in almost the same place, every time it gets to it's Pe or Ap.
Here's how you can account for that when syncing - example assumes parking orbit is lower than target.
I'll watch the target's orbit for a couple orbits - the Pe will "wander" back and forth. I'll guess where the middle of that swing is - and use that for a manual target in SyncMFD. Then I'll tune it a bit so that the manual target point is at the target's Pe - at the same time as the target reaches it's Pe.
Next, I'll note the target's altitude when it reaches Pe - remember, the PeA value is really only accurate when the vessel is at it's Pe.
Next, I'll raise my Ap to the target's altitude that I made note of, ensuring that my Ap is lined up with the Manual Target in SyncMFD.
After that, all "tuning" of my orbit is made at the intertsect (my Ap) - only my PeA get's changed.
DTMin can change because of non-spherical gravity as well as OrbitMFD - so DTMin will be most accurate when you are at the intersect - it will change during the orbit but tend to be very close to the same value each time you reach intersect again.
