Given Mehdi's latest video, I actually think both are describing different aspects of the same thing. Limited bend radius as Mehdi explains describes the resulting geometry, but it doesn't provide any reasoning about how the forces are transferred in the system to realize that geometry. In my view, I'm not convinced by Mehdi that the countering force is negligible. In most of Mould's tests it seems quite pronounced, overcoming friction and clearly moving the entire pile of chain along the floor. Add the simulations that Mould's viewers did and it becomes more convincing. Even Mehdi's own 'tubular' apparatus is clearly shaking as the chain whips back and forth, so energy transfer to/from the support is definitely going on.
Intuitively to me, it seems that where the chain/rope curves away from the 'floor', there must be a downward force because of the limited bend radius and the rope naturally 'wanting' to exceed it and move directly in the most efficient direction.
Very interesting discussion.