I have read quite a bit of this thread with interest. The article at this link is I think very relevant and an interesting read: nswagtc.org.au/files/information/gifted_acceleration_WEBSITE_long_version.pdf
It does appear to be based on research in Australia. Based on this and other articles I have read in the past, my understanding is that a good solution for the exceptionally able (the 1 in 1000 or 1 in 10,000 say) is acceleration by moving to a higher age group, and possibly even repeated acceleration for truly exceptional students. I think in the UK this only happens in private schools, and then rarely, but it makes sense, as ability grouping doesn't help children in this category. Also, since a teacher who teaches a mixed ability class of 30 every year may only see 1 such child in her entire career (and even a senior school teacher who teaches her subject to 100 children a year on average only comes across such a child once every 10 years) the teacher is unlikely to have strategies for dealing with this.
For the top 2% of the ability range, ability grouping in a fully comprehensive school can help (as the article suggests) but isn't completely the answer. Even in a school with a large year group - say 200 in the year - the top set compromises say the top 15% of the ability range. On average, there will be 4 kids out of the thirty in the 2% ability bracket. These children are as different IQ wise from the bottom child in the top set (who is still in the "above average" IQ category) as that bottom of the top set child is from the child who's IQ is exactly average. In theory though, this should not be a difficult or novel situation for the teacher - every teacher, every year, is likely to have at least 1 such child in the class.
My own school experience (too long ago and not in the UK or Australia) was of ability grouping along these lines, and some teachers were better at managing the ability range within the top set than others. It also mattered more for some subjects than others; maths being probably the most difficult to teach across a range of abilities. University was a revelation for me - not just socially in terms of meeting others like me, but in the stimulation and challenge of having a peer group of similarly able and indeed more able students and being taught at the right level for me.
Based on this experience, I have been in fortunate to be able to make different choices for my DC, who are at private selective schools. DD1 is in year 7, and very able in maths. Even at her very selective school, maths is in sets, while most other subjects aren't. The ability range in the top set is obviously much narrower since the entire year group is, say, top 15% ability wise. In her top set (pretty much all of whom will have finished primary school L6 in maths) she has a number of similarly able and even more able mathematicians. Interestingly, they only have 2 double maths lessons a week. So instead of getting the able pupils to GCSE level in maths one or more years early, the school covers what is needed each year quickly and has more time for providing teaching in other subjects - Two MFL, Latin, a current affairs class, some philosophy etc. As a result, DD is perhaps not progressing as quickly in maths as her ability suggests she could, but I am fine with that as she is also not bored and enjoying all the new subjects she is doing instead. She is moderately talented in music, sings and plays two instruments, so has plenty else to distract her. Like some others on this thread, I also think it would be counterproductive to teach her more school maths at home.
This has turned into something of an essay, sorry.