revise unit tests for ChemEnvFingerprint about matminer HOT 2 CLOSED

Hi @computron,
the CsCl structure is in fact distorted. The Cl atom/ion is not at the perfect center of the cube. And, the unit test features that (value smaller than 1 for both strategies). I significantly increased the deviation from the perfect center position and I still get the same value for the simple and multi-weight strategy. Apparently, both neighbor finding algorithms still find the same neighbors. Also, embedding in a 2x2x2 supercell didn't change the CSM values, but the computation started taking much more time. So, it doesn't seem that easy to find a case where the two differ. Do you know a small, yet very distorted structure in the MP database on top of your head?

from matminer.

Ok if you distorted the structure and don't get different values, let's just forget it.

from matminer.