Array intersection metric

Elastic Stack Elasticsearch
4

Actually, after a chat on IRC, I found a solution that is doing almost what I want. By running this query

{
  "size": 0, 
  "query": {
    "bool": {
      "filter": {
        "terms": {
          "target_ids": [
            "000",
            "111",
            "222"
          ]
        }
      }
    }
  },
  "aggs": {
    "bucket": {
      "terms": {
        "field": "ref_id",
        "size": 10
      },
      "aggs": {
        "metric": {
          "terms": {
            "field": "target_ids",
            "include": ["000", "111", "222"], 
            "size": 3
          }
        }
      }
    }
  }
}

I am only interested in the actual size of the target_ids terms aggregation. I understand that doing this from a custom application point of view is super easy (just count the key in the bucket) but I am trying . to plug this into Kibana to get back a sort of Cardinality count but only on the restricted value I passed in the filter and include.
Is there any way to get back from Elasticsearch the size of the target_ids terms aggregation instead the terms aggregation results?

I actually created a new metric in Kibana to run the restricted unique count on the passed filter but it seems that Kibana is able to handle metrics that return a single value per bucket instead of another aggregation. Another solution would be to have Kibana be able to handle responses that are not just a value but a more complex JSON.
But this is getting out of the main scope of my question.