Typo in sub_subset that leads to a crash in the isothermal sphere case. This bug has been there forever. @jregan uncovered it today by running the isothermal sphere case which has a rather deep tree. Re-assign to me if happy. See merge request !183