Index utils

In our context, the word “index” often refers to a data structure designed specifically for fast similarity lookup, given a reference population and query samples. This is especially useful for Nearest Neighbors Search (NNS). In scio, we perform such operations using the popular library faiss. The following utilities ease its use when implementing Scores relying on NNS.

Index(*, dim, metric)	Practical faiss index wrapper.
make_indexes(all_samples[, all_groups, ...])	Prepare multiple search indexes.