Run first extraction before starting webserver

It leaves less time for client to hit 503 error on startup, but more importantly, it deals with race condition in startup sequence: if extractor_task gets interrupted after webserver startup was initiated but before make_server() returned, then potoken_server.stop() becomes a no-op, leaving background task with webserver to run indefinitely.

potoken_generator/main.py

@@ -30,8 +30,8 @@ def print_token_and_exit(token_info: Optional[TokenInfo]):
 async def run(loop: asyncio.AbstractEventLoop, oneshot: bool,
               update_interval: int, bind_address: str, port: int) -> None:
     potoken_extractor = PotokenExtractor(loop, update_interval=update_interval)
+    token = await potoken_extractor.run_once()
     if oneshot:
-        token = await potoken_extractor.run_once()
         print_token_and_exit(token)
 
     extractor_task = loop.create_task(potoken_extractor.run())