Issues with normalise for polynomials.

[fingolfin]

This comes from a discussion on Nemocas/Nemo.jl#1652.

Most normalise methods in AA look like this

function normalise(a::Poly, n::Int)
   while n > 0 && iszero(a.coeffs[n])
      n -= 1
   end
   return n   # <- this is WRONG?!
end

However the documentation says normalise returns the length of the polynomial after normalisation, which is the degree PLUS ONE, which disagrees with what that code does.

In contrast, the normalise methods in Nemo seem to follow what the documentation does.

So: any code that has been using normalise probably is subtly broken: either it gives wrong results, or it is adapted to the wrong implementations of normalise. We need to figure out which it is and either adapt the documentation, or the implementations (probably both to some degree).