Evaluation of a cancer pain model for the testing of long‐acting analgesics. The effect of ms contin in a double‐blind, randomized crossover design