[ZEPPELIN-921] Apply new mechanism to KnitR and RRepl

### What is this PR for?
Applies new registration mechanism for R interpreter

### What type of PR is it?
[Refactoring]

### What is the Jira issue?
https://issues.apache.org/jira/browse/ZEPPELIN-921

### How should this be tested?
Build zeppelin with R profile enabled and run several R snippets

### Questions:
* Does the licenses files need update? No
* Is there breaking changes for older versions? No
* Does this needs documentation? No

Author: Igor Drozdov <igor_drozdov@epam.com>

Closes #1946 from DrIgor/ZEPPELIN-921 and squashes the following commits:

9ad6f28 [Igor Drozdov] New registration mechanism for R interpreter

r/src/main/java/org/apache/zeppelin/rinterpreter/KnitR.java

@@ -34,12 +34,6 @@
 public class KnitR extends Interpreter implements WrappedInterpreter {
   KnitRInterpreter intp;
 
-  static {
-    Interpreter.register("knitr", "spark", KnitR.class.getName(),
-        RInterpreter.getProps()
-    );
-  }
-
   public KnitR(Properties property, Boolean startSpark) {
     super(property);
     intp = new KnitRInterpreter(property, startSpark);

r/src/main/java/org/apache/zeppelin/rinterpreter/RRepl.java

@@ -34,12 +34,6 @@
 public class RRepl extends Interpreter implements WrappedInterpreter {
   RReplInterpreter intp;
 
-  static {
-    Interpreter.register("r", "spark", RRepl.class.getName(),
-            RInterpreter.getProps()
-    );
-  }
-
   public RRepl(Properties property, Boolean startSpark) {
     super(property);
     intp = new RReplInterpreter(property, startSpark);

r/src/main/resources/interpreter-setting.json

@@ -0,0 +1,48 @@
+[
+  {
+    "group": "spark",
+    "name": "r",
+    "className": "org.apache.zeppelin.rinterpreter.RRepl",
+    "properties": {
+      "rhadoop.cmd": {
+        "envName": "HADOOP_CMD",
+        "defaultValue": ""
+      },
+      "rhadooop.streamingjar": {
+        "envName": "HADOOP_STREAMING",
+        "defaultValue": ""
+      },
+      "rscala.debug": {
+        "envName": "RSCALA_DEBUG",
+        "defaultValue": "false"
+      },
+      "rscala.timeout": {
+        "envName": "RSCALA_TIMEOUT",
+        "defaultValue": "60"
+      }
+    }
+  },
+  {
+    "group": "spark",
+    "name": "knitr",
+    "className": "org.apache.zeppelin.rinterpreter.RRepl",
+    "properties": {
+      "rhadoop.cmd": {
+        "envName": "HADOOP_CMD",
+        "defaultValue": ""
+      },
+      "rhadooop.streamingjar": {
+        "envName": "HADOOP_STREAMING",
+        "defaultValue": ""
+      },
+      "rscala.debug": {
+        "envName": "RSCALA_DEBUG",
+        "defaultValue": "false"
+      },
+      "rscala.timeout": {
+        "envName": "RSCALA_TIMEOUT",
+        "defaultValue": "60"
+      }
+    }
+  }
+]

r/src/main/scala/org/apache/zeppelin/rinterpreter/RInterpreter.scala

@@ -109,18 +109,6 @@ object RInterpreter {
   private val logger: Logger = LoggerFactory.getLogger(getClass)
   logger.trace("logging inside the RInterpreter singleton")
 
-  // These are the additional properties we need on top of the ones provided by the spark interpreters
-  lazy val props: Map[String, InterpreterProperty] = new InterpreterPropertyBuilder()
-    .add("rhadoop.cmd",          "HADOOP_CMD", "rhadoop.cmd", "", "Usually /usr/bin/hadoop")
-    .add("rhadooop.streamingjar", "HADOOP_STREAMING", "rhadooop.streamingjar", "", "Usually /usr/lib/hadoop/contrib/streaming/hadoop-streaming-<version>.jar")
-    .add("rscala.debug",          "RSCALA_DEBUG", "rscala.debug","false", "Whether to turn on rScala debugging") // TEST:  Implemented but not tested
-    .add("rscala.timeout",        "RSCALA_TIMEOUT", "rscala.timeout","60", "Timeout for rScala") // TEST:  Implemented but not tested
-    .build
-
-  def getProps() = {
-    props
-  }
-
   // Some R interactive visualization packages insist on producing HTML that refers to javascript
   // or css by file path.  These functions are intended to load those files and embed them into the
   // HTML as Base64 encoded DataURIs.