Add support for buckets sub aggregations (split series)

[dupuyjs]

Add support for buckets sub aggregations (split series)

Provide a solution to split an existing primary aggregation using another one (for instance, following terms aggregation is split again using month date histogram, so each carrier bucket provide a count metric for each month).

  "aggs": {
    "3": {
      "terms": {
        "field": "Carrier",
        "order": {
          "_key": "desc"
        },
        "size": 5
      },
      "aggs": {
        "4": {
          "date_histogram": {
            "field": "timestamp",
            "calendar_interval": "1M",
            "time_zone": "Europe/Paris",
            "min_doc_count": 1
          }
        }
      }
    }
  },

The concept is to apply a new query on each individual bucket produced by Terms aggregation. Need some additional investigation but we can probably reuse the work done on TopHits with BuildPartitionQuery.