I think the blending to another animation needs to happen in such a way every layer creates a blend transition, instead of the other layers transitioning from their current states to the target sync state. If your diagnosis is right then this is easily accomplished in the code.
Yes. I agree.
After some further testing with the animation transition issue, I've figured out exactly what is happening.
It turns out the layers of a non active animation are not cycling as I first thought, but instead have memory of their last known active state.
Basically each layer has memory of it's last active state from when that animation was last active.
So if a layers last known state was state 3, you enter the animation again and sync to state 1 of that layer, it will first snap to it's last known state, being state 3 in this case, then transition to the synced state 1.