ID is a reserved singleton category for the row. It is used to uniquely identify the inference record. It is typically a UUID. ID columns are prefixed with the category :id: in the column name. ID columns are strings and SHOULD be unique. There is no reserved specifier or name for IDs. For this reason, :id: and :id.str: are all valid ID columns.