add "runs"

[thartbm]

For experiments with tasks run across several days, we'd not want to have to duplicate task configurations... with all the human error associated with that. So if we implement runs, we can redo the same task multiple times, provided we tell the experiment runner which "run" is being done right now.

This should be reflected in the file path.

The run number should factor into the random number generator states as well.

I'm undecided on whether or not we set the number of runs people can for each given experiment (default is 1). And if it is 1 we don't need to use it in the file path (but maybe we'd like to do that anyway?). If it is larger than 1 and we run a participant, the code can check how many runs have been completed for the participant and set the current run number (or stop it from running if the max has been hit already).