Regression test: two.pf is producing somewhat different results when run twice with the same commit

[kslong]

Although all of the other tests produce identical results when run with the same commit, two.pf does not. The results are similar but not identical, as they should be since it is not being run with a random seed. Here is what the comparison looks like in the regression plot.

(Aside: the commands.txt file for the regression tests has not been updated and still refers to python, though this probably does not matter for anything. All of the regression tests are run with the defaults. It's more generally unclear that we have a good set of tests, which is something that should be reviewed. However the problem with two.pf is independent of these concerns)

[kslong]

This problem depends on the number of threads. It does not appear with 8 threads, but does appear with 12.