Test sensitivity of different neuropil subtraction methods to artifacts

[lauraporta]

Is your feature request related to a problem? Please describe.
Neuropil subtraction can be done in different ways and it is not clear which of them is the most appropriate for our use case.

Currently using AllenSDK NeuropilSubtract class

Describe the solution you'd like
At least another method to compare to, and enough flexibility to switch between the two.
I'd like also to find a metric to rank one solution against the other.