fix(rust/sedona): Strip schema metadata when input uses RecordBatchReaderProvider

[paleolimbot]

Foreign record batch readers/arrow array streams often contain schema metadata attached by Pandas or R, and sometimes that metadata can interfere with the ability of DataFusion internals to propagate a schema in such a way that the metadata remains unchanged (and thus fails assertions for optimizer rules recreating an identical schema).

The true fix is somewhere in DataFusion; however, as a workaround we can strip schema metadata on input since we aren't using it anyway.

Closes #477

[Copilot]

Pull request overview

This PR addresses a schema metadata issue when using RecordBatchReaderProvider with foreign record batch readers (from Pandas or R). The solution strips schema metadata on input to work around DataFusion's schema propagation issues that cause optimizer rule failures.

Changes:

• Added a helper function to strip schema metadata from SchemaRef objects
• Updated RecordBatchReaderProvider constructor to strip metadata from input schemas
• Added a regression test using GeoPandas data to verify the fix

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
rust/sedona/src/record_batch_reader_provider.rs	Implements schema metadata stripping logic and updates the constructor to use it
python/sedonadb/tests/test_sjoin.py	Adds regression test for spatial joins with Pandas metadata

---

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