Scaling issues related to the smoothing length iterations

Discussion triggered by @nnrw56's question.

The maximal change in volume a particle can have over the course of a time-step is set by the parameter

SPH:
   max_volume_change:     2.

The default value is 2 if nothing is specified. That leads to a maximal change in h of 1.26 over the course of a time-step.

Setting this to a smaller value leads to particles being active more often but might make some of the time-steps cheaper.