Should we explore DiskANN for aKNN vector search?

### Description

I came across this compelling sounding [JVector project](https://foojay.io/today/jvector-1-0/) which looks to have awesome QPS performance.

It uses [DiskANN](https://www.microsoft.com/en-us/research/publication/diskann-fast-accurate-billion-point-nearest-neighbor-search-on-a-single-node/) instead of HNSW (what Lucene uses now).

Maybe we should explore another aKNN Codec implementation using this? It'd also be a good test of the Codec pluggability of our KNN implementation.

Start by reviewing Lucene's current aKNN Codec implementation and the KNN implementation referenced in the issue, then compare the DiskANN approach used by JVector. Done means establishing whether a DiskANN-based Codec is feasible and documenting how it would test KNN Codec pluggability.

Written by the indexing model from the issue text.