[FEATURE] Add support for nested multi-bucket terms aggregation.

Is your feature request related to a problem?

Currently only single-bucket nested aggregation are supported. So reading a response to a search request with nested terms aggregation seems impossible.

What solution would you like?

Add a new Aggregate kind called NestedTermsMultiBucket with the .buckets() method returning Buckets<NestedTermsBucket> and each NestedTermsBucket having a .key(), .docCount() and .aggregations() method.

What alternatives have you considered?

The only plausible alternative is just using an HTTP client and parsing a raw JSON.

Do you have any additional context?

Example request:

{
    "size": 0,
    "aggs": {
        "outer": {
            "terms": {
                "field": "application",
                "size": 9999
            },
            "aggs": {
                "inner": {
                    "terms": {
                        "field": "eventCategory",
                        "size": 9999
                    }
                }
            }
        }
    }
}

Example response:

{
    "took": 86,
    "timed_out": false,
    "_shards": {
        "total": 1,
        "successful": 1,
        "skipped": 0,
        "failed": 0
    },
    "hits": {
        "total": {
            "value": 114,
            "relation": "eq"
        },
        "max_score": null,
        "hits": []
    },
    "aggregations": {
        "outer": {
            "doc_count_error_upper_bound": 0,
            "sum_other_doc_count": 0,
            "buckets": [
                {
                    "key": "dataModel",
                    "doc_count": 66,
                    "inner": {
                        "doc_count_error_upper_bound": 0,
                        "sum_other_doc_count": 0,
                        "buckets": [
                            {
                                "key": "Success",
                                "doc_count": 66
                            }
                        ]
                    }
                },
                {
                    "key": "sed",
                    "doc_count": 35,
                    "inner": {
                        "doc_count_error_upper_bound": 0,
                        "sum_other_doc_count": 0,
                        "buckets": [
                            {
                                "key": "Success",
                                "doc_count": 32
                            },
                            {
                                "key": "Failure",
                                "doc_count": 3
                            }
                        ]
                    }
                },
                {
                    "key": "Mon",
                    "doc_count": 7,
                    "inner": {
                        "doc_count_error_upper_bound": 0,
                        "sum_other_doc_count": 0,
                        "buckets": [
                            {
                                "key": "Success",
                                "doc_count": 7
                            }
                        ]
                    }
                },
                {
                    "key": "alexandrina",
                    "doc_count": 6,
                    "inner": {
                        "doc_count_error_upper_bound": 0,
                        "sum_other_doc_count": 0,
                        "buckets": [
                            {
                                "key": "Success",
                                "doc_count": 6
                            }
                        ]
                    }
                }
            ]
        }
    }
}

Start by tracing the existing Aggregate kinds and support for single-bucket nested aggregations in the Java client. Add the requested NestedTermsMultiBucket kind with buckets() returning Buckets, and provide key(), docCount(), and aggregations() on each bucket. Done means the example nested terms response can be read through these APIs rather than raw JSON.