add input signals to build_explicit_observed_function

[baggepinnen]

The PR adds the possibility to specify input signals when building an observed function, this is useful for control systems. The resulting function signature will be

$$y = h(x, u, p, t)$$

where $u$ are the inputs.

If this looks reasonable, I can add a few tests.

[ChrisRackauckas]

Looks reasonable. Add a few tests.

[baggepinnen]

The tests are failing on v1.6 due to RGF emitting an opaque closure. I deactivated the added tests on 1.6

LoadError: The function body AST defined by this @generated function is not pure. This likely means it contains a closure or comprehension.
  Stacktrace:
    [1] (::RuntimeGeneratedFunctions.RuntimeGeneratedFunction{(Symbol("##arg#15198614511220004104"), Symbol("##arg#9481781105307072587"), Symbol("##arg#1087180715767745052"), :t), ModelingToolkit.var"#_RGF_ModTag", ModelingToolkit.var"#_RGF_ModTag", (0x01a3dfe2, 0xf37029e8, 0xc2df25d1, 0xb0e6e8c4, 0x919d8e05), Expr})(::Vector{Float64}, ::Vector{Float64}, ::Vector{Real}, ::Int64)
      @ RuntimeGeneratedFunctions ~/.julia/packages/RuntimeGeneratedFunctions/b1u1T/src/RuntimeGeneratedFunctions.jl:128
    [2] top-level scope
      @ ~/work/ModelingToolkit.jl/ModelingToolkit.jl/test/input_output_handling.jl:148

[YingboMa]

Why would it generate a closure at all?

[baggepinnen]

No idea, but it did generate a closure that accessed the input. The generated code is working correctly though.