Initial Condition Bug

[m-bossart]

There is some issue in the initialization for decomposition simulations.

For a decision model with a single time step, the initial condition that is used in the commitment constraint (https://github.com/NREL-Sienna/PowerSimulations.jl/blob/48ab31e7a2669febdbfda674f04b89e5f5610f97/src/devices_models/devices/thermal_generation.jl#L792)
does not appear to be getting updated properly. This does not happen in the same simulation setup if we use a normal PSI decision model that is not decomposed.

[m-bossart]

I am never hitting this method in my setup when I have a decomposed model which I don't understand: https://github.com/NREL-Sienna/PowerSimulations.jl/blob/48ab31e7a2669febdbfda674f04b89e5f5610f97/src/simulation/initial_condition_update_simulation.jl#L147.
I do hit it with the full model.

[m-bossart]

Figured out that the initial conditions don't get properly added to the decomposed optimization container. The subproblem containers have initial conditions, but not the top level main container of MultiOptimizationContainer

[m-bossart]

If I try to just use the initial conditions from one of the subproblems, I run into an error when reconciling the ICs across problems: