Specify the storage class when reading in the table so that it

doesn't randomly pick between astropy tables and pandas.

tests/test_schema.py

@@ -51,7 +51,7 @@ def _validateSchema(self, dataset, dataId, tableName):
         expectedColumns = {column['name']: column['datatype']
                            for column in sdmSchema[0]['columns']}
 
-        df = self.butler.get(dataset, dataId)
+        df = self.butler.get(dataset, dataId, storageClass="DataFrame")
         df.reset_index(inplace=True)
         outputColumnNames = set(df.columns.to_list())
         self.assertEqual(outputColumnNames, set(expectedColumns.keys()), f"{info} failed")