Thanks a lot for your work, but you can not reproduce my case with a different mesh (by the way: contact/target are flipped at your end). As already mentioned: I presented you a very simple model, but lately (2021 R1) I've been getting this element distortions more and more often between MPC-bonded contacts of different geometries - be it small, simple or very large models. It is random.
I get the same result for both solvers.

For further investigation I solved several models with different Pinball Radii.
Figure 1: Initial Information from the Contact Tool
Figure 2: Constraint Equations (for illustration purposes) after successful solution with Large Deflection: Off.
Figure 3: Total Deformation with Large Deflection: On.

Case 1: Pinball-Radius: 0,18888 mm - Unsuccessfull!

Case 2: Pinball-Radius: 1 mm - Unsuccessfull!

Case 3: Pinball-Radius: 0,001 mm - Successfull!

Case 1 & 2: You can clearly see that some constraints are built behind the Washer Diameter. Those are the spots where the element distortions occur.
Case 3: If I set the Pinball Radius very small, no constraint are built behind the Washer Diameter and the solution is successful.
Case 4: Another way to avoid constraints behind the Washer Diameter is by switching from "Nodal-Normal to Target" to "Nodal-Normal from Contact" . This will produce the exact same results as pictured in Case 3.
All Cases will successfully solve when Large Deflection is Off.
I can not comprehend why constraints behind the touching regions should be of any problem. Especially when I flip Contact/Target side, the constraint equations become even "wilder" regarding their spread ...
... but it solves flawlessly:
But this is not always the case! Sometimes the other side gets distorted! This was just an illustration, not an acceptable workaround for me.

Is this a bug?
Please help me to resolve & understand this issue.
This happens randomly & is not very uncommon anymore.
Did this never occur to you?
Thank you!