Extend Deterministic Equivalent policies to work with general uncertainties

[andrewrosemberg]

Currently, Deterministic Equivalent policies only work with uncertainties that have the same size as the state.

The problem is that we haven't decided how to properly take into account the initial state when inferring the policy. Currently, we just sum the initial state to the uncertainty in the first stage:

DecisionRules.jl/src/simulate_multistage.jl

Line 12 in 6aab59d

uncertainties_stage = initial_state .+ uncertainties_stage

But this is not general.

[andrewrosemberg]

done in #16