That's one of the reference documents we also use. Somehow I still feel like KDF (or "one-way KDF") describes a broader set of algorithms than we truly mean, but you make a good point that if some official body writes it this way, the terminology is probably at least correct enough.