Good analysis of the goals of this test. That said, the failure was interesting to me.
In the SpaceX feed they were waiting for the second stage to separate and light, and it didn't do either. So first question is why didn't it separate? I think that is the relevant one because I would expect them to have interlocks that would prevent the second stage from starting up if it were still attached to the booster. The interlocks would explain not starting up the second stage engines.
One anomaly I noticed was that the booster did not appear to shut down 100%, ever. If that was the case, that would be important because the onset of "zero g" from engine shutdown is a well tested way for the next stage to know that the stage below it has stopped firing.
Some (many? all?) spacecraft rely on the physics of the expended boost stage being much lighter than the remaining stack. This disparity means atmospheric drag on the lower stage has a bigger effect than it does on the upper stage. If the second stage retention system "lets go" when it detects zero-g, then the two pieces will begin to separate, slowly at first, and then faster when the very un-aerodynamic opening to the top of the booster gets into the slip stream.
Depending on how the second stage interlock works, it could be timed (wait t seconds and then light up) or radar (wait for distance between first and second stage) or some communication between the booster and the stage. By time is cheap, but risks a slowly separating booster being "too close" and having it become damaged from the blast of the vacuum engines starting up. Radar is likely most reliable (and most expensive), and comms would work by there is a lot of ionization going on at that speed that makes them more difficult.
So questions I would look at; First, if all the raptors didn't shut down, why not? Second, did the booster attempt a rotation and re-entry burn with the second stage still attached? Third, is there a way to "abort to orbit" this scenario where the second stage decides to let go and light up even though the first stage hasn't shut down?
Fun to watch though, its a pretty amazing rocket and that they didn't have 4 (or 5) of the raptors firing when it started off was another thing to look into. On pad aborts are a thing too, but as the parent states, there are faults that you won't get any useful data, and those you note and continue on through. I think the non-start of the raptors was of the latter.