Sometimes, oemof-solph results in marginal negative capacities or flows, eventhough the bounds clearly state a >=0. It happens because of the cbc algorithm, I suppose.

We need to decide if we cut those values from the results when evaluating with modules E.

I am removing all functions that until now cut all negative optimized capacities in the cost evaluation in E2. I suspect that we will run into some negative costs after merging PR #520