Potential for de-duplicating portions of assert-invariants via backend-specific protocol/s.

[moea]

Along the lines of the half-baked Backend protocol used in #3's tests - I think there may be some terseness wins.