[SPARK-10459] [SQL] Do not need to have ConvertToSafe for PythonUDF

JIRA: https://issues.apache.org/jira/browse/SPARK-10459

As mentioned in the JIRA, `PythonUDF` actually could process `UnsafeRow`.

Specially, the rows in `childResults` in `BatchPythonEvaluation` will be projected to a `MutableRow`. So I think we can enable `canProcessUnsafeRows` for `BatchPythonEvaluation` and get rid of redundant `ConvertToSafe`.

Author: Liang-Chi Hsieh <viirya@appier.com>

Closes #8616 from viirya/pyudf-unsafe.

sql/core/src/main/scala/org/apache/spark/sql/execution/pythonUDFs.scala

@@ -337,6 +337,10 @@ case class BatchPythonEvaluation(udf: PythonUDF, output: Seq[Attribute], child:
 
   def children: Seq[SparkPlan] = child :: Nil
 
+  override def outputsUnsafeRows: Boolean = false
+  override def canProcessUnsafeRows: Boolean = true
+  override def canProcessSafeRows: Boolean = true
+
   protected override def doExecute(): RDD[InternalRow] = {
     val childResults = child.execute().map(_.copy())