supports_constraint

[blegat]

supports_constraint of DualOptimizer simply redirects the same call the the underlying solver. This is incorrect.
Instead, we should have:

• supports_constraint of Union{VectorOfVariables, VectorAffineFunction}-in-K -> supports_constrained_variable-in-dual_set_type(K) (see [RFC] supports_constrained_variable MathOptInterface.jl#987)
• supports_constrained_variable in K -> supports_constraint of VectorAffineFunction-in-K

[guilhermebodin]

So let me see if I got this right, we should replace

function MOI.supports_constraint(optimizer::DualOptimizer, F::Type{<:SF}, S::Type{<:SS})
    return MOI.supports_constraint(optimizer.dual_problem.dual_model, F, S)
end

by two functions would those be supports_add_constrained_variable and supports_add_constrained_variables from this PR jump-dev/MathOptInterface.jl#1004?

[blegat]

No, we should keep supports_constraint but replace change its implementation. It should take into account that it dualizes the problem0