Qt frontend: Handle 'aborted' prompt replies.

On some rare occasions, the execution request for prompt
is aborted by the kernel (due to an error in another execution
request), which is not handled gracefully by the ipython frontend
raising `KeyError: 'execution_count'`
This commit simply handles 'aborted' prompt requests by retrying it.

IPython/qt/console/ipython_widget.py

@@ -175,11 +175,15 b' class IPythonWidget(FrontendWidget):'
         msg_id = msg['parent_header'].get('msg_id')
         info = self._request_info['execute'].get(msg_id)
         if info and info.kind == 'prompt':
-           number = msg['content']['execution_count'] + 1
-           self._show_interpreter_prompt(number)
-           self._request_info['execute'].pop(msg_id)
+            content = msg['content']
+            if content['status'] == 'aborted':
+                self._show_interpreter_prompt()
+            else:
+                number = content['execution_count'] + 1
+                self._show_interpreter_prompt(number)
+            self._request_info['execute'].pop(msg_id)
         else:
-           super(IPythonWidget, self)._handle_execute_reply(msg)
+            super(IPythonWidget, self)._handle_execute_reply(msg)
 
     def _handle_history_reply(self, msg):
         """ Implemented to handle history tail replies, which are only supported