[SPARK-2469] Use Snappy (instead of LZF) for default shuffle compression codec

This reduces shuffle compression memory usage by 3x.

Author: Reynold Xin <rxin@apache.org>

Closes #1415 from rxin/snappy and squashes the following commits:

06c1a01 [Reynold Xin] SPARK-2469: Use Snappy (instead of LZF) for default shuffle compression codec.

Author: rxin

core/src/main/scala/org/apache/spark/io/CompressionCodec.scala

@@ -56,7 +56,7 @@ private[spark] object CompressionCodec {
     ctor.newInstance(conf).asInstanceOf[CompressionCodec]
   }
 
-  val DEFAULT_COMPRESSION_CODEC = classOf[LZFCompressionCodec].getName
+  val DEFAULT_COMPRESSION_CODEC = classOf[SnappyCompressionCodec].getName
 }
 
 
@@ -103,7 +103,7 @@ class LZFCompressionCodec(conf: SparkConf) extends CompressionCodec {
 /**
  * :: DeveloperApi ::
  * Snappy implementation of [[org.apache.spark.io.CompressionCodec]].
- * Block size can be configured by spark.io.compression.snappy.block.size.
+ * Block size can be configured by `spark.io.compression.snappy.block.size`.
  *
  * Note: The wire protocol for this codec is not guaranteed to be compatible across versions
  *       of Spark. This is intended for use as an internal compression utility within a single Spark

core/src/test/scala/org/apache/spark/io/CompressionCodecSuite.scala

@@ -46,7 +46,7 @@ class CompressionCodecSuite extends FunSuite {
 
   test("default compression codec") {
     val codec = CompressionCodec.createCodec(conf)
-    assert(codec.getClass === classOf[LZFCompressionCodec])
+    assert(codec.getClass === classOf[SnappyCompressionCodec])
     testCodec(codec)
   }