Another observation is that if we ignore the existence of race AB, a person of race A with "mixed" genetics, and a person of race B with mixed genetics, may be

Another observation is that if we ignore the existence of race AB, a person of race A with "mixed" genetics, and a person of race B with mixed genetics, may be closer to eachother genetically than other members of race A and B (since as you say, they're closer to race AB, not A or B). Defining discrete categorisations for things that exist on a continuum is definitionally arbitrary and misleading. Don't tell that to psychiatrists though, who do EXACTLY what is being criticised here and slap people who exist on a continuum with an arbitrary label, and yet the DSM-V is still taken seriously by the scientific mainstream.