Add an example with hypercube constraint

[blegat]

The following is infeasible:

using DynamicPolynomials
using SumOfSquares
using MosekTools

solver = optimizer_with_attributes(Mosek.Optimizer, MOI.Silent() => false)
model = SOSModel(solver);

@polyvar x
V = 1 + x

S = @set x^2 == 1

@variable(model, γ)
@constraint(model, V <= γ, domain = S)

@objective(model, Min, γ)
optimize!(model)
@show termination_status(model)
moment_matrix(c)

It's a bit counter-intuitive and we should mention something in the docs and have an example. What happens is that the monomials of V - γ are [x, 1] so the newton polytope method find [1] which does not work. Using maxdegree solves the issue, we should mention the importance of maxdegree when there is equalities in the domain.

using DynamicPolynomials
using SumOfSquares
using MosekTools

solver = optimizer_with_attributes(Mosek.Optimizer, MOI.Silent() => false)
model = SOSModel(solver);

n = 1
@polyvar x[1:n]
V = 1 + sum(x)

S = algebraicset([x[i]^2 - 1 for i in 1:n])

@variable(model, γ)
c = @constraint(model, V <= γ, domain = S, maxdegree = 2)

@objective(model, Min, γ)
optimize!(model)
@show termination_status(model)
gram_matrix(c)

Moreover, it's unclear how to create the hypercube from a vector of x. We should have an example with algebraicset and mention something in the doc