Need to pass expressions containing unicode identifiers as split strings

since the string processing will convert the unicode sequences into regular
characters before the identifier is recognized. The fact that Monkey was
cool with this is a separate bug.

js/tests/ecma_3/Unicode/uc-003.js

@@ -43,9 +43,9 @@ function test()
                    "Escaped non-ASCII Identifier test");
     reportCompare (16, eval("++A\u03B2"),
                    "Escaped non-ASCII Identifier test");
-    reportCompare (25, eval("c\u0061se"),
+    reportCompare (25, eval("c\\u00" + "61se"),
                    "Escaped keyword Identifier test");
-    reportCompare (26, eval("++c\u0061se"),
+    reportCompare (26, eval("++c\\u00" + "61se"),
                    "Escaped keyword Identifier test");
     
     exitFunc ("test");