Adding tests for high cardinality numeric term aggregation

[vinaykpud]

Description

Adding a new test for High cardinality numeric term aggregation. This is added as part of new algorithmic changes we added to optimize numeric term aggression : opensearch-project/OpenSearch#18702

Issues Resolved

Resolves #18703

Testing

• [ Yes ] New functionality includes testing

Tested by running this tests in local

eg:

opensearch-benchmark execute-test \
--pipeline=benchmark-only \
--workload=geonames \
--target-hosts=localhost:9200 --kill-running-processes \
--include-tasks "numeric-term-cardinality-agg-high" \
--telemetry=node-stats \
--user-tag="type:numeric-after"

Backport to Branches:

• 6
• 7
• 1
• 2
• 3

---

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[gkamat]

Please indicate which versions of OpenSearch this change should be backported to.
Also, how long does this particular query take to run, roughly? If the time is significant, it may need to be enclosed in an Jinja optional block, so as not to impact users in general, but if it is fast, that is not an issue.

[vinaykpud]

@gkamat I am not sure about 'which versions of OpenSearch this change should be backported to'. But we added opensearch-project/OpenSearch#18702 in OS 3.2. Will this help us to decide for which version we need to backport?

how long does this particular query take to run, roughly?

A single query takes roughly 300 to 400 ms based on my tests for request size close to bucket size.

Also to run this tests we might need to set search.max_buckets to some value above the expected buckets as by default its only 65535. max_buckets depends on number of unique keys. In this use case its geonameid so I think depending on the maximum documents we can set the max_buckets value

[gkamat]

Backporting to the 3 branch for now. If this needs to be backported to other versions, please indicate so (for instance to compare performance against 2).