I don't think Luke, at least, is conflating morality with personal preference-optimization. He's saying: Different people have different notions of "should"-ness, and if someone says "What should I do?" then giving them a good answer has to begin with working out what notion of "should" they're working with. That applies whether "should" is being used morally or prudentially or both.

Also: What makes a moral agent a moral agent is having personal preferences that give substantial weight to moral considerations. And what such an agent is actually deciding, on any given occasion, is what serves his/her/its goals best: it's just that among the important goals are things like "doing what is right" and "not doing what is wrong". So, actually, for a moral agent "personal preference-optimization" will sometimes involve a great deal of "what morality is actually about".