Different results on DuckDBLinker and SparkLinker

[shivam221098]

When I use DuckDBLinker on Example, I get different results,

When I use the same code on SparkLinker, I get different results.

I understand that DuckDB and Spark employ distinct methods and approaches to handle data processing, which might contribute to the variations in results. However, I'm curious if there are any additional strategies or workarounds that can help ensure consistency between the deduplication outcomes obtained from DuckDBLinker and SparkLinker.