Separate between adding to batch and running the simulation

[hendrikverdonck]

In the current BladedModel.run_simulation implementation the model is added to the batch and the batch simulations are started. If multiple simulations have to be done in parallel it is assumed that the controlling python process is run in parallel (as in wtuq). It does not work well when for example running a new simulation in a for-loop, because the loop will wait until the simulation has finished. Therefore, it should be possible to just add models to the batch and then start the batch when all models are added.