When no specific context or definition is supplied, one should default to the dictionary meaning for maximum chance of successful communication. Otherwise, it is almost certain that the discussion will be about the definition or context, and thus superficial; never reaching the actual interesting stuff. It is like arguing about when to argue about a practical thing, instead of arguing about the practical thing directly.
For a resistor, this is
"an electric component that transmits current in direct proportion to the voltage across it" (
Wiktionary);
"a part of an electrical circuit designed to produce a particular amount of resistance to the flow of current" (
Cambridge dictionary);
"electrical component that opposes the flow of either direct or alternating current, [...] Resistors can have a fixed value of resistance, or they can be made variable or adjustable within a certain range, in which case they may be called rheostats, or potentiometers." (
Britannica).
In other words, by default, when one uses "resistor" without a specific context or definition, it refers to a circuit element with fixed resistance, and thus a linear relationship (direct proportion) between voltage and current over it.
None of this is about electrical engineering per se; this is purely about how to communicate with others
effectively. If you disagree, you're basically offering your own theory about effective communication between humans, so you better have a ton of proof about your assertions. A couple of eminent scientists or half-nut Nobel laureates is not enough; they are so focused on their own area of expertise that their opinions about human communications are basically irrelevant. (In mathematics, physics, et cetera, this is amply evidenced by these scientists often developing their own notations, because they fail to convey the concepts effectively using existing notations.)
Now, if one bothers to additionally compare the dictionary references to "nonlinear resistors", one will see that terms
"nonlinear element" or
"nonlinear component" are preferred and more widely used; see for example
Wikipedia,
Britannica/memristor, and so on.
Considering all of the above:
I for one fully reject the term "nonlinear resistor", and insist you use the standard term, "nonlinear element" or "nonlinear component" instead.
If we were to accept such twisted terms, we should accept "enhanced bicycle" for an automobile, "wide-ended stick" for a shovel, "multi-prong spoon" for a fork, and so on. Such terms
reduce the likelihood of concepts being conveyed without degradation, without bringing anything positive to the discussion; only allowing semantic games and trickery, which leads to discussions where nobody learns anything new. Therefore, such twisted terms should be rejected with maximum prejudice.