Because the racial example is making a direct association between skin color and behavior, the skin color being the only contributing factor which at least is scientifically unsupported and at most, ignorant and prejudice.
The sherpa example is basically giving a group of people the benefit of the doubt because of past observations, knowledge of their culture, prior experiences, and basic understanding of the value system a group of people hold and attain.
Doesnt mean the sherpa cant be wrong, but long years of interaction and experiences have given the sherpas the benefit of the doubt to many, right or wrong.
If i say bill stole my watch because he is black, there is no scientific or logical evidence to make this association.
If i say i think bill didnt steal $20 out of my wallet, even though others are accusing him, because in the past, he has asked to borrow money and always returned it, has displayed a consistent set of moral principles since i have known him, and has turned in lost items in the past on buses and trains with money intact.... I am giving giving him the benefit of the doubt based on prior experiences and a handful of observational evidence.
Theres a big difference... But in the last example, it still doesnt mean bill didnt steal my money. It just means i highly doubt it based on experience and his reputation