Test how span allocation works w/o tracing segments on 2nd-surface-only (canonical) pixels

[Algomorph]

~Potential fix for #203~

Not clear whether we need to allocate live blocks in places where only canonical surface is visible to the camera.

However, it is not clear whether significant areas like this will exist once we drop the background removal from the pipeline: after all, we would get values almost everywhere for the live surface within the camera shot.