Hello. They may look similar but there is a subtle but important difference. I will try my best to articulate it.
The "super-spine" topology was offered mostly for certain large customers who were already heavily invested in a cable plant designed for the classic three-tier approach (i.e before ACI became a thing). In those cases, the cost of recabling could end up costing more than the fabric itself. So, to give flexibility and options using the existing cable plant, they can opt for the super-spine approach. Note that this architecture assumes everything is in the same DC or within metro fiber distance where dark fiber / DWDM is readily and cheaply available (yes, I know the words 'cheap' and 'dark fiber' are almost impossible in the same sentence). The pods referenced are more to show that in large installations, the DC might be spread out over a few nearby, but different buildings. The point to make here is that there is no WAN involved.
In case of Multi-Pod and Multi-Site, the design assumes multiple DCs that are more geographically distant from each other, thus needing a WAN suitable to handle the DC interconnect. The attractive thing about how ACI does it, is this WAN can be any generic L3 routed network using equipment from any vendor. There are some basic requirements**, like a larger MTU, OSPF, DHCP Relay and in the case of Multi-Pod, PIM-BiDir. As long as those requirements are met, ACI can simply run MP-BGP as the control plane and VxLAN as the dataplane between PODs or Sites. This is a very simple and graceful, non-complex DCI and I like it very much. We can go greater distances and use networks like MPLS from our service providers.
Hope that helps
** Always read the release notes and config guide for exact details
