Fix TopHitsAggregationBuilder adding duplicate _score sort clauses

[Hohol]

When using High Level Rest Client Java API to produce search query, using AggregationBuilders.topHits("th").sort("_score", SortOrder.DESC)
caused query to contain duplicate sort clauses:

"sort": [
    {
        "_score": {
            "order": "desc"
        }
    },
    {
        "_score": {
            "order": "desc"
        }
    }
],

Fixes #42154

[elasticmachine]

Pinging @elastic/es-core-features

[elasticmachine]

Pinging @elastic/es-analytics-geo

[iverase]

Thanks for the PR! Changes look good, would you mind adding a test for this specific bug (always good to test explicitly :). After that, I can start the merging process.

[Hohol]

I added a unit test, but I'm not sure if it's good enough.
I checked what happens if we use the old version of sorting by _score in TopHitsAggregatorTests and it just causes a search to fail with exceptions. So I've made a test that simply checks if there are no exceptions.

If that's not enough, could you suggest how to test it more properly?

[iverase]

The test is enough, I will start now the CI checks and if successful I merge the PR. Thanks again!

@elasticmachine test this please

[iverase]

LGTM

[Hohol]

Looks like these CI failures are not caused by my changes.

[iverase]

Indeed, there were some problems with version bumps and a PR has just been merged to address it. May I ask you to merge master into your branch so I can restart the Jobs?

Sorry about this extra work.

[Hohol]

Done.

[iverase]

@elasticmachine test this please

[Hohol]

Again, some weird error in elasticsearch-ci/bwc.

[iverase]

@elasticmachine run elasticsearch-ci/bwc

[Hohol]

Failed again.
"stack_trace" : "ResourceNotFoundException[No datafeed with id [old-cluster-datafeed-with-aggs] exists]
Still doesn't look like caused by this PR.

[iverase]

Yes, I do not see any relation between this PR and the errors in the job. I will trigger it again and if it fails again I will look into it more in depth.

[iverase]

@elasticmachine run elasticsearch-ci/bwc

[polyfractal]

Builds have been relatively unhealthy today. There was a broken bwc test (muted here: 455cf8b) which wasn't obvious for a while because there were other errors from the version bump.

Not sure at what point you merged master in, but you'll need to get past 455cf8b at the very least. Might be easiest to just let the builds settle down a day or so and try again once they are green.

Apologies for the hassle! We can lay some of the red builds at my feet 😳

[Hohol]

Merge was before 455cf8b
I've rebased PR

[polyfractal]

<3 thanks @Hohol

@elasticmachine test this please

[iverase]

@elasticmachine run elasticsearch-ci/1
@elasticmachine run elasticsearch-ci/2

[iverase]

Thanks for the PR @Hohol, I will back port it now to 7.x.