See my reply above. I think your view is either naive, or you understand 'implementation decision' in too narrow a way.
I understand "implementation decisions" in quite well-defined way; only functional invariants should be tested. Your example is a perfect illustration for one.
Additional point is that invariant-based testing can easily be transformed into automated fuzzy tests using Quickcheck and its analogues, and this can find immense amount of otherwise hidden bugs.
> only functional invariants should be tested.
That makes no sense at all.
> Your example is a perfect illustration
I didn't give an example.
> invariant-based testing can easily be transformed into automated fuzzy tests
'fuzz' not 'fuzzy'. True, but irrelevant to the point being argued.
There shouldn't be any. Those are not useful at all.