Use MultivariatePolynomials

[blegat]

Now that MultivariatePolynomials is lightweight thanks to JuliaAlgebra/MultivariatePolynomials.jl#41, we could add it to the requirement of Polyhedra and makes the HRepresentation and Polyhedron types BasicSemialgebraicSet subtypes.
We could define a Polyhedra.Variable type with a Polyhedra.@polyvar macro and an AffineTerm and an AffinePolynomial types.
This way, one could define an HRepresentation with

@polyvar x y
@set x + y == 1 && x >= 0 && y >= 0

The inequalities and equalities iterators would return AffinePolynomials.
@rdeits what do you think ?

[chriscoey]

If this would make it easier to work with Newton polytopes for SOS optimization than I'm in favor!

[blegat]

I don't think it would help with Newton polytopes, it just gives a nice interface to build polytopes similar to SemialgebraicSets, I don't think it is a high priority.