Better behavior around ignore_missing when databases aren't available

[joegallo]

Closes #87345

The documentation for ignore_missing says:

If true and field does not exist, the processor quietly exits without modifying the document

But if the database isn't available, then we unconditionally add tags to the document regardless of the value of the ignore_missing configuration, which doesn't seem correct.

This PR aligns the behavior in the database-unavailable case to better match the behavior in the database-available case (and, by extension, to better match the docs).

[joegallo]

See especially #87345 (comment)

[elasticsearchmachine]

Pinging @elastic/es-data-management (Team:Data Management)

[elasticsearchmachine]

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

[joegallo]

Per the description on #87345, a good test for this behavior is:

POST /_ingest/pipeline/_simulate
{
  "pipeline": {
    "processors": [
      {
        "geoip": {
          "field": "client.ip",
          "target_field": "client.geo",
          "ignore_missing": true
        }
      }
    ]
  },
  "docs": [
    {
      "_index": "index",
      "_source": {}
    },
    {
      "_index": "index",
      "_source": {"client": {"ip": "8.8.8.8"}}
    }
  ]
}

Without this PR, the result of this is (note the tags on the first document):

{
  "docs" : [
    {
      "doc" : {
        "_index" : "index",
        "_version" : "-3",
        "_id" : "_id",
        "_source" : {
          "tags" : [
            "_geoip_database_unavailable_GeoLite2-City.mmdb"
          ]
        },
        "_ingest" : {
          "timestamp" : "2025-04-29T16:49:01.40054Z"
        }
      }
    },
    {
      "doc" : {
        "_index" : "index",
        "_version" : "-3",
        "_id" : "_id",
        "_source" : {
          "client" : {
            "ip" : "8.8.8.8"
          },
          "tags" : [
            "_geoip_database_unavailable_GeoLite2-City.mmdb"
          ]
        },
        "_ingest" : {
          "timestamp" : "2025-04-29T16:49:01.400909Z"
        }
      }
    }
  ]
}

But with this PR, the result is (note the lack of tags on the first document):

{
  "docs" : [
    {
      "doc" : {
        "_index" : "index",
        "_version" : "-3",
        "_id" : "_id",
        "_source" : { },
        "_ingest" : {
          "timestamp" : "2025-04-29T16:54:23.761992Z"
        }
      }
    },
    {
      "doc" : {
        "_index" : "index",
        "_version" : "-3",
        "_id" : "_id",
        "_source" : {
          "client" : {
            "ip" : "8.8.8.8"
          },
          "tags" : [
            "_geoip_database_unavailable_GeoLite2-City.mmdb"
          ]
        },
        "_ingest" : {
          "timestamp" : "2025-04-29T16:54:23.762687Z"
        }
      }
    }
  ]
}

[dakrone]

Can you also add a blurb to the docs that clarifies the behavior of ignore_missing when the database is missing?