iop.Polynomial.Evaluate should work in Lagrange/Lagrange shifted form

[ThomasPiellard]

Currently the Evaluate method on polynomial panics if the polynomial is not in canonical form.

We should store the coset in the polynomial structure (and perhaps a root of unity of the correct order) to evaluate it in Lagrange or Lagrange on a coset form. It might be useful if the polynomial is sparse in those basis for instance.