Stats aggregation: as_string fields missing when searching in multiple indices

[msh (Markus)]

Hi,

here are steps to reproduce:

• Fresh installation of ES v 8.11
• Create an index "items" with a document containing a date:

curl --location --request PUT 'localhost:9200/items/_doc/1' \
--header 'Content-Type: application/json' \
--data '{"date": "2023-01-01"}'

• Create another index "data" containing a document with a text:

curl --location --request PUT 'localhost:9200/data/_doc/1' \
--header 'Content-Type: application/json' \
--data '{"someText": "foobar"}'

• Aggregate with stats aggregation over date field in item index:

curl --location 'localhost:9200/items/_search?size=0' \
--header 'Content-Type: application/json' \
--data '{
    "aggregations": {
        "md.issued.date": {
            "stats": {
                "field": "date"
            }
        }
    }
}'

• Result:

{
    "took": 1,
    "timed_out": false,
    "_shards": {
        "total": 1,
        "successful": 1,
        "skipped": 0,
        "failed": 0
    },
    "hits": {
        "total": {
            "value": 1,
            "relation": "eq"
        },
        "max_score": null,
        "hits": []
    },
    "aggregations": {
        "md.issued.date": {
            "count": 1,
            "min": 1.6725312E12,
            "max": 1.6725312E12,
            "avg": 1.6725312E12,
            "sum": 1.6725312E12,
            "min_as_string": "2023-01-01T00:00:00.000Z",
            "max_as_string": "2023-01-01T00:00:00.000Z",
            "avg_as_string": "2023-01-01T00:00:00.000Z",
            "sum_as_string": "2023-01-01T00:00:00.000Z"
        }
    }
}

• Aggregate with stats over date field in all (*) indices:

curl --location 'localhost:9200/*/_search?size=0' \
--header 'Content-Type: application/json' \
--data '{
    "aggregations": {
        "md.issued.date": {
            "stats": {
                "field": "date"
            }
        }
    }
}'

• Response:

    "took": 1,
    "timed_out": false,
    "_shards": {
        "total": 2,
        "successful": 2,
        "skipped": 0,
        "failed": 0
    },
    "hits": {
        "total": {
            "value": 2,
            "relation": "eq"
        },
        "max_score": null,
        "hits": []
    },
    "aggregations": {
        "md.issued.date": {
            "count": 1,
            "min": 1.6725312E12,
            "max": 1.6725312E12,
            "avg": 1.6725312E12,
            "sum": 1.6725312E12
        }
    }
}

Why are the *_as_string fields (max_as_string, min_as_string, ...) missing in the response when I search over both indexes, and they are there when I search over the "items" index only? This used to work in Elasticsearch version 7.5, the fields were available in both cases.
Is this a bug or intended behaviour?

Thank you for any help and hints.