org.apache.spark.sql.execution.command
Compute stats for the given columns.
(row count, map from column name to ColumnStats) This is visible for testing.