I know that this function calculate the flash output required for a shot and store it in memory. When and why would you use this function in the real life.
Thanks,
Same reason you'd use normal exposure lock - if you were to reframe the camera would recalculate it wrongly. For example, think of a photo of a person's face in the extreme corner of the frame, when the rest of the frame is dominated by snow further away. When you initally frame the shot with the face in the centre, the camera will focus on, and meter based on, that face. It will also choose the flash output. If you didn't lock those settings, then reframed, it'd recalculate them and the face would be overexposed.
FEL is also a flash equivalent of partial (or spot) metering. While regular E-TTL flash uses some kind of matrix metering to evalate what is the correct flash exposure, when you use FEL it mesures just the area around center of the frame.
This means you can successfully use FEL to meter just a face of a performer on black background, instead of trying to light up the black and overexpose the rest.
This would be an example of such picture:
It's good for blinkers.
If the metering pre-flashes give someone in your group enough time to blink, meter the flash in advance, then you'll only flash once when you take the shot, which doesn't give them time to blink.