I'd like to comment on Martin's tale of contrasted autistic students in his maths class.
For sure, the autistic students he describes are very different from each other - there is a significant contrast between Student A who needs just a little adjustment in order to participate, and Student B who does not participate but knits and occasionally shouts.
But it is a leap of logic to automatically conclude from this that Student B is impaired.
Student B is certainly a lot more divergent than Student A: a lot more different from typical students. Classroom education is not working for Student B, and this will not be solved by a few adjustment.
But wider divergence is not the same as impairment. It may be that this student is not suited to the teaching format or maybe to the classroom environment. It may be that s/he would learn well in a very different set-up - perhaps one-to-one, perhaps an element of self-teaching, perhaps being allowed to drive their learning through their own interests and habits (there is surely scope for learning maths via knitting?).
Or it may be that this student is impaired.
My point is that we don't know for sure just from observing that Student B is more divergent than Student A.