Currently, this TYPE (rapid-failover) of L3 failover you see inthe video is currently only supported on the MSR. However, using the 175s, you could still get redundant path failover in the same way you see in the video (it would just take longer for the mesh to reconverge).
In the video's example topology, we have an AP at the North, South, East, and West. Let's call Node 1 as the west AP and we'll make it a portal. The other APs, (N/S/E) are all points. The East point will get network access thorugh the portal via either the North or South points (multi-hop mesh), so for this excercise let's call it via the South point. If we power off the south point, the East point will scan for a viable option to re-converge to, and will see the North point and re-associate and re-build the mesh. From that point on, any wireless clients or wired devices hanging behind the East point will now be able to get back to the wired network. That failover time could vary depending on RSSI, nearby mesn neighbors, etc.
Regarding points and portals, you are right, Portals are locked into being a 'gateway' devices and as such only serve as access mediums to points. Points however, can associate to portals, OR other points. How many points can connect to another point or a portal, vary depending on the configuration of your mesh profiles. In addition, any points can be configured such that their wired interface is an access port in whatever VLAN you want, or as a trunk port carrying ALL VLANs.
So right now, the main feature of the MSR suite is the advanced Layer3 routing. There's no need to file an RFE as work is currently ongoing to fold the MSR software features and capabilities into AOS, as well as AOS support for mutli-radio mesh routers (like the MSR4K) and the ability to do both 5Ghz on a 175.