I think that it is worth considering the difference between 'highly able' and 'normally able' children when discussing this. Some authors use a non-linear scale of giftedness, which is probably most helpful at these extremes - so 1 in 10, 1 in 100, 1 in 1000, 1 in 10,0000.
Approaches that work well for a '1 in 10' level for giftedness child - 3 in every class, the 'top half of the top table' in traditional terms - may well look very different for those that are needed for a 1 in 10,000 child, who are working at a level which a teacher may only see once in a career.
An approach in which a 1 in 10 giftedness child is asked to explain something to, or work with, an average child in the class, may be very successful on occasion in securing their understanding - though not as a continuous thing, and not as an alternative to proper support within the classroom.
However, the 1 in 1000 or 1 in 10,000 child - I have encountered, though not taught, one of the latter - has a different set of needs, and is sufficiently uncommon not to show up in the statistical studies predominantly made up of 'normally able' or 'moderately able' - 1 in 10, 1 in 100 - type children. After all, they are 10 or 100x rarer.
In the case of very highly able children, yes, year acceleration - or more frequently, arrangements to access work closer to their level from many years ahead in the education system - may work better, but these cases are RARE. Ibn the case of the child i know, arrangements were made early in secondary school for the child to access the sixth form maths lessons, and later to be tutored at an individual level by a university academic, though for all other subjects they were simply pushed just over the boundary (being near an age group boundary anyway) into the next yea up.
I think what I am saying is that it can be simultaneously true that for the vast majority of 'normally gifted / moderately gifted' children, year acceleration is not of benefit, and working with a slightly less able peer may work well on occasion, AND that for the much smaller group of very highly gifted children, acceleration may be one part of the solution.