Implement compability of market_optimization with disaggregations and pf

[ClaraBuettner]

The separate market_optimization was so far not used and tested for the temporal disaggregation, pf_post_lopf and spatial disaggregation.

We need to think about possible needed changes and implement them.

[ClaraBuettner]

I thought about this issue, these are my outcomes:

The I guess most critical change with regard to the pf_post_lopf and spatial and temporal disaggregation is the fact that each power plant is modelled with three generator components with the same carrier: one representing the dispatch, one representing curtailment (ramp down) and a third one representing ramp up.
The three generators have different marginal_cost and p_max_pu/p_min_pu limits.
Regarding the pf and disaggregation steps, I suggest the following things:

temporal_disaggregation
We need to keep all different generators (including the ones representing redispatch) here. We have to make sure that all marginal_cost and p_max_pu/p_min_pu limits are transferred correctly. The final amount of redispatch will be a result of the tempral_disaggregation, not of the grid_optimization anymore. We might have to consider this in final system cost calculations and constraints limiting redispatch.

pf_post_lopf
In this step, the dispatch of all generators is fixed anyhow. I guess it dosn't matter if there is one or three generators representing one power plant. The complexity is a bit higher, but the pf is quick currently.
So there are no changes urgently necessary here. But it might make sense to combine the three generators into one power plant again to also have the correct control settings.

spatial_disaggregation
The function currently runs into an error because it is not designed for having more then one generator with the same carrier at a bus. The network in the full complexity doesn't have the ramp-up/ramp-down generators. So it could get very tricky to include them in the disaggregation.
So I would prefer to combine the generators representing one power plant before this step. If we do that, no changes in the spatial disaggregation are required. We would not be able to hand over separate dispatch and redispatch time series to the distribution grids in eDisGo. I think this is not needed, but we should check first.

Summarizing: I suggest to combine the generators representing redispatch with the one representing the dispatch before the pf_post_lopf, if we don't need a differentiation between dispatch and redispatch in eDisGo.
The pf and spatial disaggregation should work correctly by then.

We still have to decide how we store the dispatch/redispatch separately in the final eTraGo results, to still be able to analyze it afterwards.