Added user impersonation fix by @mateilucianstefan

Author: velvia

notes/0.6.2.markdown

@@ -1,6 +1,6 @@
 #Scala #akka @ApacheSpark
 
-* 1 context per JVM.  This forks off a separate process for each SparkContext, and is a major change in architecture.  (@velvia, @Ankit1010, many others)
+* 1 context per JVM.  This forks off a separate process for each SparkContext, and is a major change in architecture. (EXPERIMENTAL) (@velvia, @Ankit1010, many others)
 * Update to Spark 1.6.0 (@RaafatAkkad, @apivovarov)
 * NamedObjects, an abstraction of NamedRDDs to other types of things (@TimMaltGermany / KNIME)
 * NamedBroadcasts (@CBribiescas)
@@ -11,6 +11,7 @@
 * Chunked stream encoding for Stream[_] job output (@leone17)
 * Workaround - custom error classes for context-per-jvm (@ishassan)
 * Spark Hadoop configuration in spark.context-settings.hadoop (@hntd187)
+* Support user impersonation for an already Kerberos authenticated user (@mateilucianstefan)
 * Expose MAX_DIRECT_MEMORY to adjust direct memory buffer size (@addisonj)
 * Expose JAVA_VERSION for building with say JDK8 (@stannie42)
 * Fix docker build (@stannie42)