add option to include disturbance args in generate_control_function

[baggepinnen]

This PR adds the option to include the disturbance inputs as input arguments for the generated function. This is useful when designing state estimators where the noise model is non-additive.

@AayushSabharwal the PR includes a test, which is currently failing, likely because there is a modification to wrap_array_vars required. wrap_array_vars appears long and complicated without any comments or documentation, what does this function do and can it be made easier to extend?

[baggepinnen]

The updates did not make the test pass. Looking at the generated function

julia> f[1]
RuntimeGeneratedFunction(#=in ModelingToolkit=#, #=using ModelingToolkit=#, :((ˍ₋arg1, ˍ₋arg2, ___mtkparameters___, ˍ₋arg4)->begin
          #= /home/fredrikb/.julia/packages/SymbolicUtils/jf8aQ/src/code.jl:385 =#
          #= /home/fredrikb/.julia/packages/SymbolicUtils/jf8aQ/src/code.jl:386 =#
          #= /home/fredrikb/.julia/packages/SymbolicUtils/jf8aQ/src/code.jl:387 =#
          begin
              begin
                  begin
                      begin
                          begin
                              #= /home/fredrikb/.julia/packages/SymbolicUtils/jf8aQ/src/code.jl:480 =#
                              (SymbolicUtils.Code.create_array)(typeof(ˍ₋arg1), nothing, Val{1}(), Val{(1,)}(), (+)((+)(ˍ₋arg2[1], (*)(-1, ˍ₋arg1[1])), (^)(ˍ₋arg4[1], 2)))
                          end
                      end
                  end
              end
          end
      end))

it looks like _arg4 is the disturbance_inputs, but time should have been _arg4 and the disturbance inputs should have been _arg5. Right now, time does not appear to be included among the inputs arguments

[baggepinnen]

It turns out that it works correctly with split = false, but test failing with split=true

[AayushSabharwal]

Ah, right cachesyms does that 😅. I'll push the required changes here.

[baggepinnen]

looks good now, the test failures appear to be old

[AayushSabharwal]

Yeah these failures are "expected". I have plans to fix two of them.