For GLA, @program needs to know that copying is oplus

[jpfairbanks]

MWE:

@present Mat(FreeLinearMaps) begin
    X::Ob
    Y::Ob
    Z::Ob

    f::Hom(X,Y)
    g::Hom(Y,Z)
    h::Hom(Y⊕Z, V)
end

A = @program Mat (x::X,y::Y) begin
    v = h(f(x), g(y))
    return v
end

[jpfairbanks]

@mehalter do you think you could take a look at this? How should we handle this in general? Do we need to map Julia PL features onto the GAT term constructors in general? Should we use a mechanism like the @op macro when defining the doctrine to supply this information to the program macro?

[epatters]

Since this macro generates wiring diagrams, not morphism expressions, it does not depend on otimes in any important way. In fact, it only calls otimes once (on objects) and I think that call could be removed, thus solving this problem.

Support for additive monoidal categories needs to be added in two places:

1. Wiring diagram instances: Catlab.WiringDiagrams.AlgebraicWiringDiagrams
2. Conversion of wiring diagrams to morphism epressions: Catlab.WiringDiagrams.Expressions