Catch SyntaxError from tokenizing for traceback

I don't think there's an easy way to test this, but I've verified the
fix by hand.

Closes gh-6300

IPython/core/ultratb.py

@@ -805,8 +805,10 b' class VerboseTB(TBTools):'
                     elif token_type == tokenize.NEWLINE:
                         break
 
-            except (IndexError, UnicodeDecodeError):
+            except (IndexError, UnicodeDecodeError, SyntaxError):
                 # signals exit of tokenizer
+                # SyntaxError can occur if the file is not actually Python
+                #  - see gh-6300
                 pass
             except tokenize.TokenError as msg:
                 _m = ("An unexpected error occurred while tokenizing input\n"