Determine the order of SMIRKS patterns

[bannanc]

Something the SMIRKSifier doesn't do right now is determine the order of SMIRKS patterns.
Currently, it assumes you want SMIRKS in the order of the provided clusters.
It would be a little more interesting if ChemPer could try to optimize this order.

This will not be a priority for testing ChemPer or the first publication with the December deadline. But, it should be a priority after those tests.

[bannanc]

Depending on the number of clusters, we could do this in the purely brute force way where we try moving things around, maybe just reorder randomly 5 times before adding layers?

[bannanc]

We now have more documentation on this in the preprint. However, its not a minor issue. The answer might be that ordering clusters might change depending how you are identifying those clusters.