How it was made technically? Is the game in 3d with 2 orthographic cameras on both sides of the level? So you can see and rotate the level as you build it? Or am i overthinking it?
I actually took a different approach and have 2 main components to make the effect.
Firstly, It's a single camera that i flip the x zoom on, from positive to negative (this is what causes the sprites to flip)
Next, I have 3 tilemap layers that i adjust the collision and draw order on. When I'm flipped, the 3rd row's collision activates and gets drawn over the others, while the 1st row gets disabled and drawn under the others.
Sprinkle in a background that changes on flip and collision areas that prevent flip and you got yourself the illusion of depth.