Update _decode method to accept integer element and convert it to sequence

#### Why did I create this pull request?
When I tried to run this model over TGI, during model warm up step it fails by throwing `cannot convert 'int' to 'sequence'`.
#### What are the proposed changes?
Removed `List[int]` type mentioned in `_decode` function parameter and explicitly added a check to convert single element `token_ids` into list.
#### How did I test it?
I did changes over local repo in my machine and was able to successfully launch the model using TGI.

tokenization_xgen.py

@@ -169,7 +169,9 @@ class XgenTokenizer(PreTrainedTokenizer):
         """Converts an index (integer) in a token (str) using the vocab."""
         return self.encoder.decode_single_token_bytes(index).decode("utf-8")
 
-    def _decode(self, token_ids: List[int], skip_special_tokens: bool = False, **kwargs):
+    def _decode(self, token_ids, skip_special_tokens: bool = False, **kwargs):
+        if not isinstance(token_ids, list):
+            token_ids = [token_ids]
         if skip_special_tokens:
             token_ids = [t for t in token_ids if t not in self.all_special_ids]
         return self.encoder.decode(token_ids)