Airfoilmod
I fully agree with your assertion that a core vulnerability in ETOPS redundancy is that the top-level systems (i.e. the aircraft, external nav-comm environment, crew) are unitary and therefore not able to be much more redundant.
As you observe, a variant of the same theme is that the recent history of the aircraft is also common-mode for even the redundant systems, esp the power train. When the two halves of the aircraft are somehow identical, certain unanticipated failure modes related to patterns of usage may affect them equally at roughly identical points in time.
Perhaps the logical extension of this observation - especially if the principle eventually is shown to be causative in the case of BA038 - would be a future requirement to ensure that the two halves of an ETOPS aircraft may use many common components but must be deliberately asymmetrical in regard to the form, fit, and function of sustaining flight-critical systems. A logistics nightmare, of course, but probably a reasonable extension of the no-fail-all concept.
Nature handles this problem quite neatly in many living things, including ourselves, by making the redundant critical appendages, sensors, and control systems from the same general parts and pieces, but different in right-left (etc.) symmetry. A substantially trickier design asymmetry might be needed to cover the many plausible common-mode failure symmetries possible with aircraft.