Wrap ES KNN queries with PatienceKNN query

[tteofili]

This exposes an option to add the patience-based early termination defined in apache/lucene#14094 to KNN queries in ES.
This leverages PatienceKnnVectorQuery which wraps ES KNN queries when mapping is one of hnsw, int8_hnsw, int4_hnsw, bbq_hnsw and early_termination is set to true (within index_options).

e.g.:

{
  "mappings": {
    "properties": {
      "my_vector" : {
        "type": "dense_vector",
        "dims" : 10,
        "index" : true,
        "index_options" : {
          "type" : "bbq_hnsw"
          "early_termination" : true
        }
      }
    }
  }
}

Currently this has the limitation that we can't profile that query because we can't extend the PatienceKnnQuery same as https://github.com/elastic/elasticsearch/blob/main/server/src/main/java/org/elasticsearch/search/vectors/ESKnnByteVectorQuery.java#L28-L38 because of visibility issues. However it's fixed in 10.3 branch via apache/lucene#14838 and we'll incorporate that as soon as 10.3 is released and included in ES main branch.

[tteofili]

dense_vector

[tteofili]

so_vector

[github-actions]

🔍 Preview links for changed docs:

• docs/reference/elasticsearch/index-settings/index-modules.md

🔔 The preview site may take up to 3 minutes to finish building. These links will become live once it completes.

[elasticsearchmachine]

Pinging @elastic/es-search-relevance (Team:Search Relevance)

[elasticsearchmachine]

Hi @tteofili, I've created a changelog YAML for you.

[benwtrent]

Whew, way cleaner! Still some room for simplifying. But the index setting looks much nicer to me :)

[benwtrent]

So clean! I like it, and the performance numbers are ⚡