Description:

This PR adds SQL transpilation support for pre-aggregates to handle cross-database compatibility when the source warehouse uses a different SQL dialect than DuckDB.

Key Changes:

• Added @polyglot-sql/sdk dependency for SQL dialect transpilation between different database systems (Snowflake, BigQuery, Postgres, etc. → DuckDB)

• Implemented transpileExploreSqlFilters function that:

  • Transpiles sqlWhere expressions from source warehouse dialects to DuckDB syntax
  • Renames column references to flattened pre-aggregate column names (e.g., status → orders_status)
  • Preserves Lightdash user attribute placeholders like ${lightdash.attributes.X} and ${ld.attr.X}
  • Handles complex SQL expressions including date functions, casts, and array operations

• Enhanced pre-aggregate matching to allow matches when base tables have sqlWhere filters, since they can now be properly transpiled at query time

• Updated demo configuration with a comprehensive sql_filter example showing attribute-based filtering with comma-separated values and PostgreSQL-specific syntax

• Added comprehensive test coverage for various SQL transpilation scenarios including cross-table joins, complex date filters, and edge cases

The transpilation ensures that pre-aggregates work seamlessly across different warehouse types while maintaining the same filtering logic and user attribute support.