Yes, fatigue is the issue, but if it was caused by tensile stress, there is no reason why the cracks would always appear at the point of minimum curvature and not at the connectors (weakest section) or everywhere randomly, hence my assumption of flexural fatigue.
Try to imagine you pulling a rope against a friend: there is only tensile stress, sooner or later you or your friend just lose grip on the rope, because it is the weakest connection point. In the Macbook the connector is a “clip-on” type. If tensile stress was the problem, the cable would just slip out of the socket slowly with time. If the connector wasn’t the weakest point, the cable would break at a random point, not always the same for every Mac.
An idealized cable only has tensile stress, but as thin as it can be, a real cable still has a third dimension (thickness), so it is more correct to apply beam theory to understand the issue. Real cables can and do sustain flexural stress, otherwise you would feel zero reaction when you push one of them perpendicularly.
Actually this is the very reason “flex cables” are made as thin as possibile: to minimize the stresses induced by the bending (beam theory here again). But these stresses become zero only for zero thickness, which is impossibile.
P.s. The flex cable doesn’t have to resist bending, it should oppose the least resistance. If it had enough rigidity to resist bending, it simply wouldn’t do its job.
