'They' didn't invent torque before there was a need to. No one said "I know, let's take Force times Perpendicular Distance and call it Torque"
What was found was that long levers need less force when applied against short levers. (Obvs, of course) An empirical rule was postulated and then confirmed by countless measurements and observations. The principle of Moments was born. Torque was the word which described the effect of levers.
I don't think there can be any actual proof of this - any attempt would involving just re-stating an idea and then re-stating it in a different way - proving nothing. It's just like Momentum (another quantity that we have found to be always conserved (when you get rid of all the practicalities of an experiment)
In the case of many so-called Machines, the geometry of the thing gives a theoretical 'Velocity Ratio' and that's often confused (even PF is incredibly sloppy about this) with 'Mechanical Advantage', which is the measured ratio of load one effort. It always falls short of VR because of Inefficiency. People ignore the dead weight of lever arms, friction and the weight of pulley blocks. The MA of that Capstan in the above picture will be hopelessly worse than the VR might suggest.