[FEA] Digital Fingerprinting Updates: Explainability

[jarmak-nv]

Top 3* contributing fields

[jarmak-nv]

@gbatmaz - Please add a general high-level description here, thanks!

[gbatmaz]

Description:
-Along with the anomaly score, we will return the features that contribute to the loss most. (For each data type)
-We may also return supporting information, e.g. if those features already had high variation in the benign set etc.
I shared the code for the first part with @efajardo-nv . The second part is still pending on me.

[gbatmaz]

I think we are ok for the next release right? @efajardo-nv
We are still discussing the longer term features