Allow Bucket Script aggregation to reference on null

[larrycinnabar]

related to #36642

related to #31920

Example:

  "some-items":{  
    "aggregations":{  
      "averaged":{  
        "avg":"field"
      }
    },
    "children":{  
      "type":"child"
    }
  },
  "some-metric":{  
    "bucket_script":{  
      "buckets_path":{  
        "a":"some-items>averaged.value"
      },
      "script":{  
        "source":"params.a + 100"
      }
    }
  }

If, in this example, number of child elements are zero, the AvgAggregation can't be performed (as averaging of zero elements is impossible). That averaged aggregation will return null.
But this null will silentry break the some-metric. In the final result, some-metric won't just appear (without any error).

I think it's a good thing to be able to handle these case inside the script. For example:

"source":"if (params.a == null) {return 100} else {return params.a + 100}"

In case if this null is not handled - elastic can just return an error

[elasticmachine]

Pinging @elastic/es-analytics-geo

[elasticmachine]

Pinging @elastic/es-core-infra

[polyfractal]

Yep, this is definitely a bug. I'm going to close though since it's being tracked in #27377, and in particular there's a plan that was discussed here: #27377 (comment)

The tl;dr: is that we would like to start passing the raw value to the script, but also pass in the doc_count. Both are important to provide, so that the user knows if the value is actually null, or if it's just null because there were no documents.

Thanks for opening the issue, and sorry you're running into this :( There are several related pipeline issues that revolve around missing docs / null values, so hopefully we can start tackling them soon.