Modelingtoolkitize doesn't work if parameters are kept in a named tuple or struct

[pcjentsch]

I know this isn't supposed to work right now, but it would be nice to have!

example:

using OrdinaryDiffEq
using ModelingToolkit
p = (a = 1.01,)
f(u,p,t) = p.a*u
u0 = 1/2
tspan = (0.0,1.0)
prob = ODEProblem(f,u0,tspan,p)
mtk_prob = modelingtoolkitize(prob)

gives

ERROR: LoadError: KeyError: key 'a' not found

and with structs

using OrdinaryDiffEq
using ModelingToolkit
using Parameters
@with_kw struct parameter_struct
    a = 1.01
end
p = parameter_struct()
f(u,p,t) = p.a*u
u0 = 1/2
tspan = (0.0,1.0)
prob = ODEProblem(f,u0,tspan,p)
mtk_prob = modelingtoolkitize(prob)

ERROR: LoadError: MethodError: no method matching keys(::parameter_struct)
Closest candidates are:
  keys(::Core.SimpleVector) at essentials.jl:605
  keys(::Cmd) at process.jl:639  
  keys(::LibGit2.GitTree) at C:\Users\peter\.julia\packages\Revise\AMRie\src\git.jl:52
  ...

[ChrisRackauckas]

Hey! I read this and it's a great idea. It's a lower priority so I'm not going to address it right now, but sometime in the future I hope to come back to it because it would be nice to have.