It is done for a finite resistance - in this case that is dictated largely by the supply impedance. A higher resistance fault would seem unlikely given the circumstances (short run of tails between CUs - exposed wiring etc).
I just had a look at 543.1.3, which describes the application of the adiabatic equation for checking the sizing of a main earth conductor in these cases. Something I missed was that for an main earth conductor that is a "single" and not part if a cable, and not bunched with other wires, they permit a k factor of 143, rather than the normal 115 for a PVC clad wire. If one applies that here is makes the required size estimate lower (7.3mm)
For tails <= 16mm^2 the earth should be the same size as the tails. For
16 to 35mm^2 it should be 16mm^2, and over 35mm^2, half of the tails CSA.Yup, I would not lose sleep over it!