Better aggregation for date facets

Date facets currently create the same params as other facets, eg:

Array
(
    [localgov_news_categories] => Array
        (
            [terms] => Array
                (
                    [field] => localgov_news_categories
                    [size] => 500
                )

        )

    [localgov_news_date] => Array
        (
            [terms] => Array
                (
                    [field] => localgov_news_date
                    [size] => 10000
                )

        )
)

In this example, localgov_news_categories is a facet on a taxonomy field. localgov_news_date is a facet on a date field.

The response from OS for our date facet is:

Array
(
    [localgov_news_date] => Array
    (
        0 => Array
        (
            [count] => 6
            [filter] => "1647604800"
        )
        // Another 518 records removed here.
    )
)

As you can see, that search returned 519 facet values, but the eventual display only has 61 options. This is because the response returns the stored unix timestamp, which is then further aggregated by facets into monthly options.

While this approach works, it means we're moving more data than we need to. It also means the limit option in the facet doesn't make a lot of sense. The current maximum limit of 500 reduces the number of options, as there's 519 unique days in the response, even though the facet when unlimited shows a maximum of 61 months.

Using the date histogram aggregation and passing the granularity from the facet settings should fix this, I think: https://opensearch.org/docs/latest/aggregations/bucket/date-histogram/