feat(es|ql): add dense_vector support in coalesce

[mromaios]

Closes #139928

Details:
This PR adds dense_vector support to COALSESCE.

Limitations:

• The provided arguments need to be of type dense_vector, so passing in an array would require a to_dense_vector() function. I will address Implicit casting in a follow up PR

Note to reviewer:

Example requests:

DELETE my-index

PUT my-index
{
  "mappings": {
    "properties": {
      "fine_category_emb": {
        "type": "dense_vector",
        "dims": 3
      },
        "broader_category_emb": {
        "type": "dense_vector",
        "dims": 3
      },
      "my_text" : {
        "type" : "keyword"
      }
    }
  }
}

PUT my-index/_doc/1
{
  "my_text" : "text1",
  "fine_category_emb": [0.5, 10, 6],
  "broader_category_emb": [0.7, 15, 2]
}
PUT my-index/_doc/2
{
  "my_text" : "text2",
  "fine_category_emb": null,
  "broader_category_emb": [-0.7, 15, 2]
}

PUT my-index/_doc/3
{
    "my_text" : "text3",
    "fine_category_emb" : null,
    "broader_category_emb" : null
}

POST /_query
{
"query": "FROM my-index | eval search_emb = COALESCE(fine_category_emb, broader_category_emb, to_dense_vector([0.0, 0.0, 0.54])) "
}

Result:

{
    "took": 22,
    "is_partial": false,
    "completion_time_in_millis": 1771951486365,
    "documents_found": 3,
    "values_loaded": 12,
    "start_time_in_millis": 1771951486343,
    "expiration_time_in_millis": 1772383486308,
    "columns": [
        {
            "name": "broader_category_emb",
            "type": "dense_vector"
        },
        {
            "name": "fine_category_emb",
            "type": "dense_vector"
        },
        {
            "name": "my_text",
            "type": "keyword"
        },
        {
            "name": "search_emb",
            "type": "dense_vector"
        }
    ],
    "values": [
        [
            [
                0.7,
                15,
                2
            ],
            [
                0.5,
                10,
                5.9999995
            ],
            "text1",
            [
                0.5,
                10,
                5.9999995
            ]
        ],
        [
            [
                -0.7,
                15,
                2
            ],
            null,
            "text2",
            [
                -0.7,
                15,
                2
            ]
        ],
        [
            null,
            null,
            "text3",
            [
                0,
                0,
                0.54
            ]
        ]
    ]
}

[mromaios]

❓ 💭 Do we need coalesce tests in the individual .csv-spec files, or is that an overkill? (same comment for the other types (default, bit, byte))

[github-actions]

ℹ️ Important: Docs version tagging

👋 Thanks for updating the docs! Just a friendly reminder that our docs are now cumulative. This means all 9.x versions are documented on the same page and published off of the main branch, instead of creating separate pages for each minor version.

We use applies_to tags to mark version-specific features and changes.

Expand for a quick overview

When to use applies_to tags:

✅ At the page level to indicate which products/deployments the content applies to (mandatory)
✅ When features change state (e.g. preview, ga) in a specific version
✅ When availability differs across deployments and environments

What NOT to do:

❌ Don't remove or replace information that applies to an older version
❌ Don't add new information that applies to a specific version without an applies_to tag
❌ Don't forget that applies_to tags can be used at the page, section, and inline level

🤔 Need help?

• Check out the cumulative docs guidelines
• Reach out in the #docs Slack channel

[elasticsearchmachine]

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

[elasticsearchmachine]

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

[mromaios]

Check: https://github.com/elastic/elasticsearch/pull/142974/changes#r2848341088
A question about whether we want coalesce tests in the dense_vector*.csv-spec tests.

[carlosdelest]

LGTM, great job @mromaios ! 👏

One thing we're not doing is validating that we are always returning dense_vectors of the same dimensions 🤔 . That's not something we can validate at the analyzer level (as we don't have dimension information there), and tricky to do at runtime (we would have to change how the ExpressionEvaluator work).

Let's keep it this way - we could add an Evaluator that validates dimensions as a follow up PR if needed

[ioanatia]

looks good.

@carlosdelest I am not sure COALESCE needs to check whether the vectors have the same dimensions.

The documentation of COALESCE simply states:

Returns the first of its arguments that is not null. If all arguments are null, it returns null.

I worry we would overcomplicate the definition of COALESCE if we add a validation that is specific to dense_vector.
So what we have right now with this PR should suffice IMO.

[carlosdelest]

I worry we would overcomplicate the definition of COALESCE if we add a validation that is specific to dense_vector.
So what we have right now with this PR should suffice IMO.

@ioanatia agreed - I was just thinking about how COALESCE allows users to have vector elements with different dimensions, something that the mapping guarantees us not having.

I already approved and agreed to keep it this way - makes no sense to complicate this and we can always add some validation layer afterwards.