Description
Today, we can't resolve a metadata column from a Spark dataframe from Scala with df.col("metadataColName"). This is because metadata output is only used during resolveChildren (used during SQL resolution); it should likely also be used by resolve (used during Scala resolution).