Tamara Vardo’s answer is most of the answer.
I think there’s a psychological component as well, though this is getting into speculation. It’s convenient for implicature to have xor on a scale before and, and to require the less natural notion of inclusive or to be expressed as a combination of the two, rather than allow it to be implicit.
But there is a notion underlying this, that xor is a more natural notion than inclusive or, to begin with. And that’s likely to do with humans making sense of the world through binary opposition: the notion that things are either X or Y, but not both, is very helpful if you’re trying to classify the world. The notion of things being both X and Y does not really challenge a model of the world through binary opposition: you’re just introducing a new binary opposition.
But I think the notion that things may be either X or Y, and you don’t care whether they’re both or not—which is what inclusive or means—undermines binary classification. Which is why it’s not the default.
See also: Which natural language differentiates exclusive and inclusive or?