All nodes that carry a cost should be accounted for in the end. Need investigation.

We want to provide a cost breakdown per data stream accounting for the cost of the data nodes but also provide insight into how much the other nodes/data transfers is costing per cluster. E.g an ML job is costing x ECU