Glenn,
I like the way you got at that. I agree. Locknuts inside and outside of an enclosure wall.
About the bonding of the nipple at both ends. . .The raceway isn't a conductor. I agree that, when the raceway is conductive, it will be a parallel path for the unbalance current that is not returning towards the transformer through the GEC. But the neutral is chosen and sized for its current carrying properties, not the raceway except, if it is conductive, that it provide a fault current path of sufficiently low impedance as to clear the poco's OCP at the xformer.
While bonding the nipple at both ends provides even better performance, bonding it only at one end is all that is required on the poco side of the service disco & OCP.
Our poco doesn't make specific bonding requirements, rather, it leaves that to the NEC and the AHJ.