NateTG makes a very good point when he says, "You're assuming that the torque rating (which should be a time-average) is the same as the instantaneous torque at ignition.
You're also ignoring that the torque rating is the net torque and that while one cylinder is firing, other cylinders are getting compressed, going through power strokes, or venting exhaust. (The L91 is a 4-cylinder engine.)"
If you could see the instantaneous torque output from an IC engine, you would observe that there is considerable pulsation in it. This is the reason that IC engines are notorious for exciting torsional vibration in the machinery trains that they drive. The rated torque is most definitely an average torque, averaged over all cylinders and over a full thermodynamic cycle which is two crank revolutions for a 4-stroke engine.
In a multicylinder engine, when one cylinder is on the power stroke, another cylinder is on the compression stroke, so the net output torque of the engine is reduced from that of the firing cylinder by the amount required to compress the second cylinder (assuming just two cylinders for simplicity). Furthermore, there are many losses in an IC engine that also reduce the output torque including bearing friction (the main bearings are heavily loaded during part of the cycle!) and power required to drive the cam shaft and distributor, water pump, oil pump, fan, generator, etc. All of these so called "parasitic loads" reduce the available output shaft torque because they come straight off the crank shaft before any torque is delivered to the outside world.
