Common sense dictates that the "buttonhook" back to the runway is usually not a good idea.
To pull it off, you need to immediately recognize the problem, then instantly guess right about how your glide ratio, wing loading in the turn, distance from the threshold, and wind will affect such a plan. You also need to be sure that you're high enough to even contemplate it. And oh yeah, there may be traffic departing or landing behind you on the same runway. Even in a motorglider like the one in the video (which is optimized for, um, gliding), there are those other factors to consider.
Because of all this, most attempts don't work out well. Pilot takes too long to accept that the thrust is fading if not gone already, sits there slack-jawed with the plane still pitched and trimmed to climb at takeoff power, airplane slows down and may even start to descend even though the wing isn't fully stalled (yest). Next mistake is to pull back in order to claw for altitude, thus slowing the plane down more (and increasing the rate of descent), then (if the stall doesn't fully develop by then) desperately trying to "yank and bank" it back towards the runway... the result is very predictable to any pilot, but easy to forget when you're all "oh sh*t, the engine stopped!" Sadly, most pilots who've tried it have apparently forgotten all of the above, out of sheer panic.
So unless you're at some minimum altitude that you know will allow a safe turn around in whatever you're flying, and you have pitched for Vbg immediately, and you know the runway is clear and the wind won't cause a dangerous overshoot, you're better off not even considering the "impossible turn". The odds of having all those things lined up, and being aware of them in the very limited time available, are very slim.
Which is not to say you must not change heading at all... a shallow turn might be required to make a safe emergency landing with minimum damage and/or injury. You also might be able to use whatever thrust you have, if any, to help you get to such a place- a gamble, for sure, but obstacles or terrain might make an immediate descent unsafe. Whatever you decide, you have to avoid the temptation to pull the nose up to try to gain altitude. It's amazing how many pilots forget this.
The best practice is to just first pitch for Vbg, then look for a place you can see will work, based on what you've seen on any given final approach, especially a simulated engine-out. Sounds kinda like a typical engine-out training exercise done in cruising flight, doesn't it? That's because an engine-out is an engine-out, regardless of what mode of flight you're in, or where you are in relation to a runway.
If a downwind landing on the runway you just left falls within the category of "place you can see will work", then by all means turn around, keeping in mind that in a bank, the indicated stall speed will go up slightly. Remember also that in order for you to see that it will work, you have to know your altitude is sufficient, know the wind is not too strong, and know the runway and departure path are clear. I know that sounds impossible... that's why they call it the "impossible turn".
