Use Matryoshka + Binary Quantisation Embeddings

Ragatouille supports using models for encoding trained on Matryoshka embeddings but I wasn't able to find a way to use them for indexing. Secondly although ragatouille supports storing compressed vectors to disk but it would be great to combine them with binary/int 8 quantised embeddings. Would love to contribute if it's feasible.

No files or tests are named. Start by locating the indexing path and the existing compressed-vector storage implementation, then determine how Matryoshka dimensions and binary or int8 quantisation would flow through them. Done means both embedding approaches are supported for indexing and their stored-vector behavior is covered by tests.

Written by the indexing model from the issue text.