All branches I'm currently working on either need no experiments or already have 
experiments using the old MyExperiment wrapper. Therefore, I included the new 
common_setup.py in this branch. The branch has been merged and pushed.

Is there a point in having experiments for this?
I would say: if it works on one task, it's fine.

I see the point of improving common_setup.py, but shouldn't we do this in the
context of a task that actually needs experiments?

I opened a pull request at 
<https://bitbucket.org/jendrikseipp/downward/pull-request/13/issue422-memory-
logging-and-common_setuppy/diff>.

I'll use this issue for improving the experiment scripts.

It would be nice to have a more regular update on peak memory usage during
search, e.g. in the "progress evaluator" output.