Add support for has_parent, has_child and parent_id queries

[csh97]

Implements #116

Adds support for remaining join queries: has_parent, has_child and parent_id (https://www.elastic.co/guide/en/elasticsearch/reference/current/joining-queries.html)

Changes to note:

• Moved nested query into join-queries file as Elasticsearch documentation groups it under join queries
• Moved ScoreMode enum from inside nested query to join-queries file as has_parent and has_child use the same score modes
• Changed filter source extension functions to extend on JsonDsl rather than SearchDsl so that they can be used for inner hits in parent and child queries
• Added documentation for join queries (Left nested query as a TODO as I don't have any experience using nested queries)

has_parent

SearchDSL().apply {
        query = hasParent("parent") {
            query = matchAll()
            minChildren = 1
            maxChildren = 2
            scoreMode = ScoreMode.max
            ignoreUnmapped = true
            innerHits {
                filterSource("name")
                innerHitSize = 1
                name = "adfa"
                sort { add("a") }
            }
        }
    }.json(pretty = true).let { println(it) }

Produces:

{
  "query": {
    "has_parent": {
      "parent_type": "parent",
      "query": {
        "match_all": {
        }
      },
      "min_children": 1,
      "max_children": 2,
      "score_mode": "max",
      "ignore_unmapped": true,
      "inner_hits": {
        "_source": [
          "name"
        ],
        "size": 1,
        "name": "adfa",
        "sort": [
          {
            "a": {
              "order": "DESC"
            }
          }
        ]
      }
    }
  }
}

has_child

SearchDSL().apply {
        query = hasChild("child") {
            query = matchAll()
            minChildren = 1
            maxChildren = 2
            scoreMode = ScoreMode.max
            ignoreUnmapped = true
            innerHits {
                filterSource("name")
                innerHitSize = 1
                name = "adfa"
                sort { add("a") }
            }
        }
    }.json(pretty = true).let { println(it) }

Produces:

{
  "query": {
    "has_child": {
      "type": "child",
      "query": {
        "match_all": {
        }
      },
      "min_children": 1,
      "max_children": 2,
      "score_mode": "max",
      "ignore_unmapped": true,
      "inner_hits": {
        "_source": [
          "name"
        ],
        "size": 1,
        "name": "adfa",
        "sort": [
          {
            "a": {
              "order": "DESC"
            }
          }
        ]
      }
    }
  }
}

parent_id

SearchDSL().apply {
        query = parentId(type = "child", id = "123") {
            ignoreUnmapped = true
        }
    }.json(pretty = true).let { println(it) }

Produces:

{
  "query": {
    "parent_id": {
      "type": "child",
      "id": "123",
      "ignore_unmapped": true
    }
  }
}

[barbulescu]

minor - direct call can be used here and below

Suggested change

	block.invoke(q)
	block(q)

[csh97]

I used invoke() as it seems like most of the project is using it over direct call and I wanted to keep with convention of the project. But I am also happy to change to direct call

[jillesvangurp]

Nice. Also thanks for documenting this.