So here is what I infer from this discussion:
The "ethnic breakdown" in a person's results is not a literal reflection of that person's ancestry. Rather, it describes how that individual's unique admixture of specific genetic markers that reflect deep ancestry compares to a series of reference populations. Those reference populations have been selected by population geneticists with specific research goals in mind. The labels given to these populations are to a large extent arbitrary and do not necessarily refer to current countries or geographic entities. When a percentage breakdown is given for an individual, it indicates which of these reference populations have the greatest similarity to the person's admixture. The best admixture tool to choose will vary with each individual and will depend on how closely the reference populations in that tool mirror the person's actual ethnic mix.