Clarify on_ready triggering more than twice.

On that note, fix the background_task example to use `Client.run`
instead of manual loop management. That way the task will be cleaned
up and cancelled inside `Client.run`.

docs/api.rst

@@ -108,6 +108,9 @@ to handle it, which defaults to print a traceback and ignore the exception.
     .. warning::
 
         This function is not guaranteed to be the first event called.
+        Likewise, this function is **not** guaranteed to only be called
+        once. This library implements reconnection logic and thus will
+        end up calling this event whenever a RESUME request fails.
 
 .. function:: on_resumed()
 

examples/background_task.py

@@ -19,13 +19,5 @@ async def on_ready():
     print(client.user.id)
     print('------')
 
-loop = asyncio.get_event_loop()
+client.loop.create_task(my_background_task())
-
+client.run('token')
-try:
-    loop.create_task(my_background_task())
-    loop.run_until_complete(client.login('token'))
-    loop.run_until_complete(client.connect())
-except Exception:
-    loop.run_until_complete(client.close())
-finally:
-    loop.close()