Floating point equality in tests

[stsievert]

When I ran the tests, a couple tests failed because of floating point equality. The numbers were almost the same, but very slightly different. e.g., in test_msd_ratio

    assert ft.msd_ratio(dframe, 1, 9) == 0.09708430006771959
#   assert 0.09708430006771954 == 0.09708430006771959

This is only present in test_features.py in

• test_msd_ratio
• test_alpha_calc
• test_gyration_tensor
• test_boundness
• test_efficiency

In all of these, static floating point values are on one side of the equality check ==.

[stsievert]

NumPy's allclose tests to make sure that two floating point numbers (or arrays) are very close:

>>> np.allclose([1e10,1e-8], [1.00001e10,1e-9])
True