You might enjoy the fact that some experts call it the "fruit salad of autism/ADHD".
The say that spectrum is inaccurate and the fruit salad is a better name) analogy/description.
Like with fruit salad, you can serve it to a table of people and everyone will have fruit salad on their plate, but it will be randomly varied for all. Some will have a lot of one fruit and a few others. Some will have all but one and so on.
I would also say that dyslexia isn't a single general condition. But in general group of issues that affect textual communication. Person could for example have certain repeating type of issues while writing, but still be able to read well and without issues. This is also a type of dyslexia.
So, spectral (as in fuzzy bands of clumped outputs that may overlap, but may have gaps, and may in some cases be less fuzzy), but not a spectrum (a continuous, fairly smoothly distributed shape over a wide range).
That would also show consistent benefits in controlled studies.
No, not in practise. Its hard to controll for something that isn't understood enough. You would have to have a good ekough sampling across the groups, and esch group would have to be big enough.
So something may help type 1 dyslexia, but not help type 2 or type 3 etc.