Use \uXXXX escapes instead of UTF-encoded characters.

Writing them this way (from http://www.fileformat.info/info/unicode/char/2603/index.htm and http://www.fileformat.info/info/unicode/char/1f42a/index.htm) ensures that the test code will be parsed correctly no matter what charset Java is using.

This allows us to open-source the test of \u{XXXX} escapes, so our external users will surely be delighted that they can now fill their JavaScript programs with emoji, confident that the transpilation works as expected.
-------------
Created by MOE: http://code.google.com/p/moe-java
MOE_MIGRATED_REVID=90732596

test/com/google/javascript/jscomp/Es6ToEs3ConverterTest.java

@@ -1905,4 +1905,8 @@ public void testTaggedTemplateLiteral() {
     ));
   }
 
+  public void testUnicodeEscapes() {
+    test("var \\u{73} = \'\\u{2603}\'", "var s = \'\u2603\'");  // ☃
+    test("var \\u{63} = \'\\u{1f42a}\'", "var c = \'\uD83D\uDC2A\'");  // 🐪
+  }
 }