I've reduced the example to this - flick the lever and the same behavior can be observed:
[media]One theory I have about the bottom contraption transitioning from screenshot 2 to 3 is the following: The middle piston is directly powered by the observer behind it. The piston below it on the other hand is powered by quasi-connectivity. Under normal circumstances (when there's no block update above that piston), it wouldn't extend, but since there is a piston above it - the middle piston that is - it receives a block update and is thus powered and can extend. The same is true for the retraction: the bottom piston only retracts because the piston above it causes a block update.
Knowing that, and assuming that the order of operations is (1) middle piston retracts and (2) bottom piston retracts, then it would make sense that the slime block cannot be moved back, since it sticks to the block below it which itself is blocked by the immovable extended piston.
I think this might actually explain it. Let me know what you think.
I've added one made in the newest vanilla 1.16.3 snapshot.
Would you be so kind and reopen this issue? I'm not going to open a new one with the very same content.
I have also tried this with the newest Snapshot (status 2020/10/04), same results.
Should be added to the attachments.
Assuming my theory is true, the question would be if this is intended behavior or not, since you'd (or I would) normally think that the order of operations shouldn't matter when the pistons retract during the same tick.