How could an accurate simulation be created of unknown situations unless the AB test pilots intentionally put the aircraft into those situations, which, in my opinion, might lead to vacancies in the test pilot ranks?
Aircraft design engineers provide their best estimates of the limits (maximum airspeed, stall speeds in various configurations, bank angles, g-loading, etc.). The manufacturer's test pilots, who are usually also have engineering degrees, then approach these limits slowly, being alert for bothersome (squirrely) behavior. Then they make a judgment as to how close the typical line or fleet pilot should be allowed to come to the limit. How close varies with the type of aircraft. Then the user's (airline/military) own test pilots verify the manufacturer's proposed limits and decide what additional limitations, if any, should be imposed on line or fleet pilots. Aerodynamically precise simulators are designed by extrapolating the experiences of test pilots so, of course, they can't be perfect and can only be expected to predict aircraft behavior within certain limits. In the past 50 years, very few test pilots have bought the farm. That only happens when they rush the program.