That source is a bit suspicious, to say the least.
First off, coyōtl is not an ‘Aztec’ word, it is a Nāhuatl word. Aztec is a group of languages, of which Nāhuatl is the largest dialect subgroup; the other major subgroup is Pipil (also called Nawat), but in Pipil, a coyote is cuyut, not coyōtl—i.e., the word was borrowed specifically from Nāhuatl (Classical Nāhuatl, as it happens), not from just any Aztec language.
In Nāhuatl, in coyōtl means first and foremost ‘coyote’. As with many animal names, however, the word has some extended meanings, too, amongst them ‘trickster’, ‘mestizo’, and ‘sallow [yellowish]’. Generally not very positive words. (Compare the different mental images you get if you refer to a person as a dog, a snake, a cow, an ass, or a sheep.)
The extension in meaning happened in Nāhuatl, though, and the word was borrowed from Nāhuatl into Spanish with two meanings: ‘prairie wolf’ and ‘trickster’. Therefore, the Spanish term coyote is a loanword in both senses—it is not a calque in either sense. If, for example, in Otomi the term for the animal, miñ’o, had also been used to refer to a small child who is constantly howling, and the Spanish had borrowed that meaning, but using the by now already existing Spanish word coyote to translate it, that would have been a calque. Simply borrowing two meanings of a word from the source language, though, does not a calque make.
Since English has continued the borrowing and simply borrowed both meanings from Spanish, the word is not a calque in either sense in English, either. I am not aware, though, whether the narrowing down from ‘trickster’ to ‘human trafficker across the border from Mexico to the US’ is an independent development in English, or whether it occurred in Mexican Spanish and was then applied to the English word afterwards. If it is the latter, an argument could be made that English used the existing English word ‘coyote’ to create a calque of the Spanish word coyote—it would perhaps be a bit of a stretch, though.