Added feature to move added tokens in vocabulary for Transformer-XL (#4953)

* Fixed resize_token_embeddings for transfo_xl model

* Fixed resize_token_embeddings for transfo_xl.

Added custom methods to TransfoXLPreTrainedModel for resizing layers of
the AdaptiveEmbedding.

* Updated docstring

* Fixed resizinhg cutoffs; added check for new size of embedding layer.

* Added test for resize_token_embeddings

* Fixed code quality

* Fixed unchanged cutoffs in model.config

* Added feature to move added tokens in tokenizer.

* Fixed code quality

* Added feature to move added tokens in tokenizer.

* Fixed code quality

* Fixed docstring, renamed sym to 	oken.

Co-authored-by: Rafael Weingartner <rweingartner.its-b2015@fh-salzburg.ac.at>

tests/test_tokenization_transfo_xl.py

@@ -82,3 +82,16 @@ class TransfoXLTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
         self.assertListEqual(
             tokenizer.tokenize(" \tHeLLo ! how  \n Are yoU ?  "), ["HeLLo", "!", "how", "Are", "yoU", "?"]
         )
+
+    def test_move_added_token(self):
+        tokenizer = self.get_tokenizer()
+        original_len = len(tokenizer)
+
+        tokenizer.add_tokens(["new1", "new2"])
+        tokenizer.move_added_token("new1", 1)
+
+        # Check that moved token is not copied (duplicate)
+        self.assertEqual(len(tokenizer), original_len + 2)
+        # Check that token is moved to specified id
+        self.assertEqual(tokenizer.encode("new1"), [1])
+        self.assertEqual(tokenizer.decode([1]), "new1")