Saddlepoint Approximations: A Review and Some New Applications