typedef-ai/fenic

AI Summary: Implement a `count_distinct` aggregate function in the Fenic DataFrame library. This involves creating a new expression class, extending the expression transpiler to handle it (converting to a Polars expression), adding a user-facing API function, and writing unit tests.

AI Summary: The task involves adding support for the PySpark `array_join` function to the Fenic DataFrame library. This requires creating a new expression class, implementing transpilation logic to Polars, exposing a user-facing API function, and writing unit tests. The implementation needs to handle null values gracefully, potentially requiring the composition of multiple Polars expressions.