Investigate forecasting algorithm to explain 'missed' cycles

[mabijkerk]

While updating the forecasting algorithm in #1343, we paid extra attention to its functionality. I noticed some periods where I would expect the forecasting algorithm to perform a charge/discharge cycle, but it doesn't: hence the term 'missed' cycles.

This scenario (blank nl2019, 22 GW offshore wind, 25 GW solar PV, 6 GW large-scale batteries with 4 hours volume and forecasting on), produces the following behaviour:

It is clear that the hours in the end of the 18th and the beginning of the 16th are matched as a charge/discharge cycle by the algorithm. What I wonder is why the intermediate hours are not selected for another charge/discharge cycle:

@noracato could you investigate this at some point?