feat(rust/sedona-expr): Add GroupsAccumulator to framework and implementation for ST_Envelope_Agg

[paleolimbot]

This PR allows aggregate functions implemented using the SedonaAccumulator to implement their own GroupsAccumulator (which they should usually do for performance on tiny groups). This also implements one for ST_Envelope_Agg() and the appropriate extension to the AggregateUdfTester (since I couldn't seem to trigger all the features of the GroupsAccumulator reliably with simple queries).

This would also benefit other aggregate functions (but they would require that implementation to exist and it might take a bit).

Closes #167.

[Copilot]

Pull request overview

This PR implements GroupsAccumulator support for aggregate functions in the Sedona Rust framework, specifically implementing it for ST_Envelope_Agg(). The GroupsAccumulator is a performance optimization in DataFusion for efficiently aggregating many small groups.

Changes:

• Extended the SedonaAccumulator trait with optional groups_accumulator_supported() and groups_accumulator() methods
• Implemented BoundsGroupsAccumulator2D for ST_Envelope_Agg() to optimize grouped aggregations
• Enhanced AggregateUdfTester with aggregate_groups() method to test groups accumulator functionality

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
rust/sedona-testing/src/testers.rs	Added aggregate_groups() method and refactored shared schema/expression initialization
rust/sedona-functions/src/st_envelope_agg.rs	Implemented BoundsGroupsAccumulator2D and added comprehensive grouped aggregation tests
rust/sedona-expr/src/aggregate_udf.rs	Added groups_accumulator_supported() and create_groups_accumulator() to the UDF framework
python/sedonadb/tests/functions/test_aggregate.py	Added integration tests for ST_Envelope_Agg including grouped aggregations

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[jiayuasu]

Did you see any performance gain from this?

[paleolimbot]

For this particular example (ST_Envelope) it's hugely beneficial:

import sedona.db

sd = sedona.db.connect()

# 10 million points
sd.sql("""SELECT * FROM sd_random_geometry('{"target_rows": 10000000}')""").to_memtable().to_view("pts")

# pip install --pre --force-reinstall sedonadb --extra-index-url=https://repo.fury.io/sedona-nightlies/
%time sd.sql("""SELECT ST_Envelope_Agg(geometry) as e FROM pts GROUP BY id % 1000""").execute()
%time sd.sql("""SELECT ST_Envelope_Agg(geometry) as e FROM pts GROUP BY id % 10000""").execute()
%time sd.sql("""SELECT ST_Envelope_Agg(geometry) as e FROM pts GROUP BY id % 100000""").execute()
#> CPU times: user 885 ms, sys: 29.9 ms, total: 915 ms
#> Wall time: 95.7 ms
#> CPU times: user 1.03 s, sys: 13.5 ms, total: 1.04 s
#> Wall time: 99.2 ms
#> CPU times: user 15.8 s, sys: 146 ms, total: 15.9 s
#> Wall time: 1.44 s

# pip install python/sedonadb
%time sd.sql("""SELECT ST_Envelope_Agg(geometry) as e FROM pts GROUP BY id % 1000""").execute()
%time sd.sql("""SELECT ST_Envelope_Agg(geometry) as e FROM pts GROUP BY id % 10000""").execute()
%time sd.sql("""SELECT ST_Envelope_Agg(geometry) as e FROM pts GROUP BY id % 100000""").execute()
#> CPU times: user 283 ms, sys: 34.6 ms, total: 317 ms
#> Wall time: 43.1 ms
#> CPU times: user 238 ms, sys: 7.81 ms, total: 246 ms
#> Wall time: 25.7 ms
#> CPU times: user 407 ms, sys: 58 ms, total: 465 ms
#> Wall time: 44.5 ms

[jiayuasu]

Great work!