Evaluation of induced responses, insect population growth, and host‐plant fitness may change the outcome of tests of the preference‐performance hypothesis: a case study