I think using n == 0 is fine when you know for sure that n is an ordinary (and perhaps real) number. You know this because it’s an internal variable, or there is a type constraing on an input variable, for example.
iszero(x) is better when you don’t have full control over the type of x, or if it’s not an ordinary number, otherwise it seems a bit like overkill to me (though a pretty gentle overkill at that.)