Add a "new-yarn" directory in SPARK_HOME, intended to contain Hadoop-2.2 API changes.

new-yarn/pom.xml

...cala/org/apache/spark/deploy/yarn/ApplicationMaster.scala

...apache/spark/deploy/yarn/ApplicationMasterArguments.scala

.../src/main/scala/org/apache/spark/deploy/yarn/Client.scala

.../scala/org/apache/spark/deploy/yarn/ClientArguments.scala

...che/spark/deploy/yarn/ClientDistributedCacheManager.scala

...n/scala/org/apache/spark/deploy/yarn/WorkerRunnable.scala

.../org/apache/spark/deploy/yarn/YarnAllocationHandler.scala

...la/org/apache/spark/deploy/yarn/YarnSparkHadoopUtil.scala

...apache/spark/scheduler/cluster/YarnClusterScheduler.scala

...park/deploy/yarn/ClientDistributedCacheManagerSuite.scala