EXAMINE/LOOK AT very often gives you a lot more information than simply what the player character sees. A lot of the time it adds the character's memories: "This is the shirt you wore to Jessica's recital a year ago, to the disastrous Sidewalk Chalk Tournament two years ago, to the department picnic in whichever year it was you went to the department picnic. It's brown." Or, especially when the object of the verb is involved in a puzzle, its "description" includes some guesswork by the player character: "From what you can tell, when the generator was in operation, it supplied power to the nearby alarm system."
People add information from other senses in descriptions all the time, and without causing any consternation—I've yet to hear a player complain about a LOOK AT response that mentioned a smell. On the other hand, if you do mention a smell, chances are good that the player will try to SMELL the object next, and there'll definitely be some consternation if that doesn't get a response.
But if you're writing about a dog (or similarly-blessed being), I think it's a great idea to merge SMELL with EXAMINE. It makes real-life sense: A dog probably smells everything passively all the time, without any more effort than we use when we look at something. And when players notice olfactory sense-data in every EXAMINE response, and they notice that SMELL returns the exact same responses, they'll catch on very quickly that you're saving them the work of typing both commands. Plus it would be immersive!