Floating point precision

[eirrgang]

For tests like

ensemble_md/ensemble_md/tests/test_ensemble_EXE.py

Line 411 in 69de773

assert list(g_vec_2) == [0.0, 2.200968785917372, 3.9980269151210854, 3.5951633659351256, 4.897041830662871, 5.881054277773005] # noqa: E501

you are likely to need an approximate comparison. Aside from ordinary small differences in floating point arithmetic between systems or run time conditions (particularly through vectorization), complex calculations like MD results can be particularly non-deterministic through parallelization schemes.