FP - there is merit in training for it.
The most difficult thing to train for is the judgement required to decide whether the manouevre is viable or not... this is best achieved with scenario-based exercises as you suggest.
The main pinch points are the underpinning knowledge of aerodynamics required (ie high AoA flight, stalled stick position, best glide performance (ie AoA more so than speed) & the nexus between them), followed by the specific aircraft handling skills required (ability to select a control position that causes CLmax for the gliding / low power turn, the appropriate AoB & the immediate transition from high AoA flight to best glide once the turn is completed). Make no mistake, the aircraft handling required during a turnback must be "on point". Additionally, the maturity to cease the manouevre during the execution if it becomes apparent that it is not going to be successful - roll wings level & retain control to landing.
Like all things, with training & practice it can be understood to the point where it can become one of the tools legitimately available in the event of a power loss after take-off. It can be safely trained for. However, it is not the only solution, nor is it advisable in all situations - the poor decision-making or handling is where it goes wrong. As an industry, we set people up for failure on this when we don't train for it & then at some point they give it a go when it needs to count...
For mine, I am still here today as a result of successfully executing a turnback following a complete engine failure after take-off over a populated area.