Start coupling with all simulation active

[IshaanDesai]

The rationale for starting with all simulation in an active state is that if the user has provided only micro data to compute the similarity on, in the first step only one simulation will be solved and the Micro Manager will not get the actual micro-scale data to compute the adaptivity from. Therefore it is necessary to solve all the micro simulations in the first step to generate the necessary micro data from which the similarity will be computed. This also results from the fact that we no longer provide the initialize() method in the micro simulation API. As there is no initialization which the Micro Manager can call, it cannot know the micro-scale data before computing the first time step. So either we reintroduce the initialize method and initialize all micro simulations, or we solve all of them in the first time step.

Note that this problem occurs only when micro data is provided to the adaptivity. If macro data is provided, the adaptivity will be computing correctly.

[IshaanDesai]

After some discussion, it was concluded that providing only micro data for the adaptivity does not make logical sense. This perspective was brought by @mathiskelm. The adaptivity state at a particular instance will remain the same if only micro data is fed into the similarity computation. Macro data must be provided too. With this perspective, it is not necessary to solve all the micro simulations in the first time step.