[SQL] Add average overflow test case from apache#978

By @egraldlo.

Author: egraldlo <egraldlo@gmail.com>
Author: Michael Armbrust <michael@databricks.com>

Closes apache#1033 from marmbrus/pr/978 and squashes the following commits:

e228c5e [Michael Armbrust] Remove "test".
762aeaf [Michael Armbrust] Remove unneeded rule. More descriptive name for test table.
d414cd7 [egraldlo] fommatting issues
1153f75 [egraldlo] do best to avoid overflowing in function avg().

sql/core/src/test/scala/org/apache/spark/sql/SQLQuerySuite.scala

@@ -136,6 +136,12 @@ class SQLQuerySuite extends QueryTest {
       2.0)
   }
 
+  test("average overflow") {
+    checkAnswer(
+      sql("SELECT AVG(a),b FROM largeAndSmallInts group by b"),
+      Seq((2147483645.0,1),(2.0,2)))
+  }
+
   test("count") {
     checkAnswer(
       sql("SELECT COUNT(*) FROM testData2"),

sql/core/src/test/scala/org/apache/spark/sql/TestData.scala

@@ -30,6 +30,17 @@ object TestData {
     (1 to 100).map(i => TestData(i, i.toString)))
   testData.registerAsTable("testData")
 
+  case class LargeAndSmallInts(a: Int, b: Int)
+  val largeAndSmallInts: SchemaRDD =
+    TestSQLContext.sparkContext.parallelize(
+      LargeAndSmallInts(2147483644, 1) ::
+      LargeAndSmallInts(1, 2) ::
+      LargeAndSmallInts(2147483645, 1) ::
+      LargeAndSmallInts(2, 2) ::
+      LargeAndSmallInts(2147483646, 1) ::
+      LargeAndSmallInts(3, 2) :: Nil)
+  largeAndSmallInts.registerAsTable("largeAndSmallInts")
+
   case class TestData2(a: Int, b: Int)
   val testData2: SchemaRDD =
     TestSQLContext.sparkContext.parallelize(