[SPARK-8458] [SQL] Don't strip scheme part of output path when writing ORC files

`Path.toUri.getPath` strips scheme part of output path (from `file:///foo` to `/foo`), which causes ORC data source only writes to the file system configured in Hadoop configuration. Should use `Path.toString` instead.

Author: Cheng Lian <lian@databricks.com>

Closes #6892 from liancheng/spark-8458 and squashes the following commits:

87f8199 [Cheng Lian] Don't strip scheme of output path when writing ORC files

sql/hive/src/main/scala/org/apache/spark/sql/hive/orc/OrcRelation.scala

@@ -111,7 +111,7 @@ private[orc] class OrcOutputWriter(
     new OrcOutputFormat().getRecordWriter(
       new Path(path, filename).getFileSystem(conf),
       conf.asInstanceOf[JobConf],
-      new Path(path, filename).toUri.getPath,
+      new Path(path, filename).toString,
       Reporter.NULL
     ).asInstanceOf[RecordWriter[NullWritable, Writable]]
   }