wait_minutes(position, agents, avg_handle_min) estimating a caller's wait: with agents working in parallel and position callers ahead, the caller starts after ceil(position / agents) service rounds of avg_handle_min each. Position 0 waits 0. Use math.ceil.wait_minutes(7, 3, 4)
12
ceil(7/3)=3 rounds × 4 minutes.
Rounds, not per-caller minutes, is the model.
math.ceil(position / agents).