ES|QL - dense_vector approximate nearest neighbour search support

[carlosdelest]

Description

As dense_vector support is being addressed in #125783, ES|QL will be able to perform approximate nearest neighbour vector search.

The following tasks will be needed:

• kNN search function using vectors: ES|QL - kNN function initial support #127322
• ESQL - Add K mandatory param for KNN function #129763
• Prefiltering in kNN search ESQL - KNN function uses prefilters when pushed down to Lucene #131004
• kNN uses LIMIT for setting k, switches to exact search when it can't be pushed down: ESQL - KNN function uses LIMIT for K, transforms to exact search when not pushed down #132944
  • Previous work:
    • ESQL - KNN function uses LIMIT for setting top k #129353
    • ESQL - KNN functions with non-pushed down filters #131708
• ES|QL: Add TEXT_EMBEDDING function #131022
• ES|QL: Add support for semantic_text in knn function #132066
• Allow query parameters to be used in vector functions, so users can use query parameters for specifying query vectors: ES|QL - Allow multivalued query parameters #134317
• ES|QL: automatic prefiltering for semantic_text match queries #132068

[elasticsearchmachine]

Pinging @elastic/es-analytical-engine (Team:Analytics)