restart vs. continue study

[cskiraly]

For longer parameter studies it is nice if an interrupted study can be continued.
What's needed:

• allow to specify execID externally
• option to skip run if results are already there

This would also allow to change the config, e.g. adding more runs for statistical relevance, or exploring another dimension of the simulation space.

[leobago]

I like this, very nice idea.

One small thing we have to be careful is that with parallel execution, the simulation space is not traversed/finished in order, so at restart we cannot assume that because the run Y=4 is in the results directory, then all runs for y<4 are already done. We have to check one by one.