Reward for learning strategies is calculated at the wrong place

[nick-harder]

Right now the calculate_reward function is executed during set_dispatch_plan call, but this function does not reflect the dynamics of the units, these are calculated in the execute_current_dispatch function of each unit. For example for storage units, the reward is calculated before the SOC is actually updated in the execute_current_dispatch function, which leads to wrong calculations. I believe the calculate_reward as well as calculate_cashflow functions should be removed to the execute_current_dispatch functions, so we calculate the cashflow and the reward correctly when everything has been updated. @maurerle @kim-mskw what do you think?