[SPARK-4296][SQL] Trims aliases when resolving and checking aggregate expressions

I believe that SPARK-4296 has been fixed by 3684fd21. I am adding tests based #3910 (change the udf to HiveUDF instead).

Author: Yin Huai <yhuai@databricks.com>
Author: Cheng Lian <lian@databricks.com>

Closes #4010 from yhuai/SPARK-4296-yin and squashes the following commits:

6343800 [Yin Huai] Merge remote-tracking branch 'upstream/master' into SPARK-4296-yin
6cfadd2 [Yin Huai] Actually, this issue has been fixed by 3684fd21.
d42b707 [Yin Huai] Update comment.
8b3a274 [Yin Huai] Since expressions in grouping expressions can have aliases, which can be used by the outer query block,     revert this change.
443538d [Cheng Lian] Trims aliases when resolving and checking aggregate expressions

sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/SQLQuerySuite.scala

@@ -267,4 +267,19 @@ class SQLQuerySuite extends QueryTest {
     sql("DROP TABLE nullValuesInInnerComplexTypes")
     dropTempTable("testTable")
   }
+
+  test("SPARK-4296 Grouping field with Hive UDF as sub expression") {
+    val rdd = sparkContext.makeRDD( """{"a": "str", "b":"1", "c":"1970-01-01 00:00:00"}""" :: Nil)
+    jsonRDD(rdd).registerTempTable("data")
+    checkAnswer(
+      sql("SELECT concat(a, '-', b), year(c) FROM data GROUP BY concat(a, '-', b), year(c)"),
+      Row("str-1", 1970))
+
+    dropTempTable("data")
+
+    jsonRDD(rdd).registerTempTable("data")
+    checkAnswer(sql("SELECT year(c) + 1 FROM data GROUP BY year(c) + 1"), Row(1971))
+
+    dropTempTable("data")
+  }
 }