Hi team, what's the recommended approach to filter big tables that are partitioned by date? in my sources.yml:

meta:
          elementary:
            timestamp_column: "ingestion_timestamp"
            where_expression: DATE_DIFF('day', DATE(date), CURRENT_TIMESTAMP(6)) <= 3
        tests:
          - elementary.freshness_anomalies
          - elementary.volume_anomalies

Is still timing out, I can't understand the compiled queries for this, though I see the where expressions. Same where_expression works for me as the

filter

in dbt-core's

source freshness

blocks