Rework the parser so we can use arbitrary start nodes of the syntax.

This also includes a rework for error recovery in the parser. This is now just possible for file_input parsing, which means for full files.
Includes also a refactoring of the tokenizer. No more do we have to add an additional newline, because it now works correctly (removes certain confusion.

jedi/api/__init__.py

@@ -322,15 +322,12 @@ class Script(object):
 
     @memoize_default()
     def _get_under_cursor_stmt(self, cursor_txt, start_pos=None):
-        tokenizer = source_tokens(cursor_txt)
-        r = Parser(self._grammar, cursor_txt, tokenizer=tokenizer)
-        try:
-            # Take the last statement available that is not an endmarker.
-            # And because it's a simple_stmt, we need to get the first child.
-            stmt = r.module.children[-2].children[0]
-        except (AttributeError, IndexError):
+        node = Parser(self._grammar, cursor_txt, 'eval_input').get_parsed_node()
+        if node is None:
             return None
 
+        stmt = node.children[0]
+
         user_stmt = self._parser.user_stmt()
         if user_stmt is None:
             # Set the start_pos to a pseudo position, that doesn't exist but