[CI] XPackRestIT test {p0=ml/sparse_vector_search/Check sparse_vector should prune by default} failing

[elasticsearchmachine]

Build Scans:

• elasticsearch-periodic-platform-support #11279 / rocky-9_platform-support-unix
• elasticsearch-periodic-platform-support #11161 / sles-15_platform-support-unix

Reproduction Line:

./gradlew ":x-pack:plugin:yamlRestTest" --tests "org.elasticsearch.xpack.test.rest.XPackRestIT.test {p0=ml/sparse_vector_search/Check sparse_vector should prune by default}" -Dtests.seed=71F86033DB6D1873 -Dtests.locale=ms-ID -Dtests.timezone=Pacific/Gambier -Druntime.java=25

Applicable branches:
9.2

Reproduces locally?:
N/A

Failure History:
See dashboard

Failure Message:

java.lang.AssertionError: Failure at [ml/sparse_vector_search:970]: 
Expected: <2>
     but: was <12>

Issue Reasons:

• [9.2] 2 failures in test test {p0=ml/sparse_vector_search/Check sparse_vector should prune by default} (0.2% fail rate in 816 executions)
• [9.2] 2 failures in pipeline elasticsearch-periodic-platform-support (9.5% fail rate in 21 executions)

Note:
This issue was created using new test triage automation. Please report issues or feedback to es-delivery.

[elasticsearchmachine]

Pinging @elastic/es-search-relevance (Team:Search Relevance)

[john-wagster]

I haven't been able to reproduce this despite trying a few things including running it off main, 9.2 branch, and locally with about 1000 different seeds.

I tried reviewing how if at all this could possibly be failing intermittently and didn't immediately see anything.

I noticed there's some other related issues that traditionally ml-core has handled so I'm going to redirect. Ya'll feel free to kick it back if you believe that's an error on my part.
#136061

[elasticsearchmachine]

Pinging @elastic/ml-core (Team:ML)

[elasticsearchmachine]

Pinging @elastic/search-relevance (Team:Search - Relevance)

[kderusso]

@markjhoy I wonder if this is as simple as missing a teardown? I see we had to explicitly delete other indices as part of a teardown here, and you're already creating the index with a single shard.