Spent a lot of time in nuclear power school computing energy transference and losses. Electrical, thermal, radiation, mechanical, potential energy even mass. They are all forms of power. Spend a few months working steam tables and you will see what I mean.
It is easy to compute how much power is dissipated by the light bulb because it is very easy to measure the electrical power being consumed by the bulb. What ever the bulb doesn't dissipate in light energy, it dissipates as heat. So put the bulb in an insulated box, measure the heat loss, then you now know how much power is beign radiated as light. So yes, the light output can be measured by observing the electrical consumption less the thermal losses.
An engine reved against a trans brake can be producing the same HP as one powering a moveing car, yet because the car is not moving some people would say no work is being done. I look at that and know that with the trans brake on, the power is being dissipated as heat in the transmission. Release the brake and the power that was formerly heating up the trans fluid is now accelerating the car. The same amount of energy is being disspated either way, just in different forms.
Pick up an engineering text on automotive design. In a good one you will find an equation relating the maximum acceleration potential of a car to engine power. The analysis should include the effect of gearing. What you will find is two things: the ability of a car to accelerate decreases with vehicle speed and the maximum acceleration is related to the engine's mechanical power. Not torque. Torque is actually somewhat nebuluous because it can assume practically any value at the rear wheels depending upon what gear ratios are used. However, there is a limit what you can do with gearing because at a given vehicle speed you can't shift too low or it will over rev the engine.