Bug Fix for correct tracking of current investment stage by SDDP

[filippopecci]

The value of mysetup["CurrStage"] is modified in run_genx_case_multistage! inside a for loop. The dictionary mysetup is then passed to function run_ddp as setup_d and the current stage value is used by function initialize_cost_to_go but it is not updated to match the current stage, and left set to the last stage as done by parent function run_genx_case_multistage!.

This commit fixes that and set the right current stage in setup_d before initializing the cost-to-go.

[gmantegna]

Looks good

[RuaridhMacd]

Just for future reference, the bug was caused by the following:

1. In run_genx_case_multistage! the code looks over time steps t in 1:mysetup["MultiStageSettingsDict"]["NumStages"] to create the different models and inputs and store them in a dictionary.
2. In doing so, it sets mysetup["MultiStageSettingsDict"]["CurStage"] = t to pass t to generate_mode
3. That means t = mysetup["MultiStageSettingsDict"]["NumStages"] at this point.
4. run_genx_case_multistage! then calls run_ddp on the dicts of models and inputs but a shared setup dictionary
5. run_ddp then loops over the stages using t again, calling initialize_cost_to_go(shared_settings, model[t], inputs[t] for each time step t.
6. To calculate the discount factor, initialize_cost_to_go defines the time (aka cur_stage) as cur_stage = settings_d["CurStage"], which is the total number of stages, based on steps 1 -> 3