Fix parser from SDDP

[andrewrosemberg]

Currently, the parser (a series of code scripts) only grabs the first problem from the sddp.jl graph.

How I currently bring problems from sddp.jl:

• I first save the file related to the first problem:

  DecisionRules.jl/examples/HydroPowerModels/run_sddp.jl

  Lines 40 to 44 in 5f413f5

  	# ## Save subproblem
  	# results = HydroPowerModels.simulate(m, 2);
  	# model = m.forward_graph[1].subproblem
  	# delete(model, all_variables(model)[findfirst(x -> x == "", name.(all_variables(model)))])
  	# JuMP.write_to_file(model, joinpath(case_dir, string(formulation)) * ".mof.json")

• Then, by knowing the name of the variables that my sddp.jl builder creates and how sddp.jl names incoming and outgoing stages, I transform the variables that represent parameters to the choice of parameter representation:

  • Context sddp.jl: parameters there are just fixed variables (jump name for variable in equality constraint) - so I need to find the variable and the constraint (to delete it if needed or save the reference).
  • DecisionRules.jl will soon only allow for parameters to be just MOI.Parameter.
  • code: https://github.com/LearningToOptimize/DecisionRules.jl/blob/main/examples/HydroPowerModels/load_hydropowermodels.jl