> But the relation of those groups with what we call "race" is almost zero.

The everyday races correspond to real, observable genetic clusters. There are real characteristics shared by members of these large groups, e.g. the epicanthic fold in East Asians.

The genetic phylogeny doesn't lie. The Sub-Saharan diversity you're mentioning --- e.g., between Bushmen and the Igbo --- is about large-separation clades within the larger continental grouping.