Bug: raising NackMessage() ignores the retry parameter

[sidjakinsboriss]

Describe the bug
If NackMessage is raised in the message handler, the retry parameter is ignored, and the messages keeps getting re-queued

How to reproduce
Include source code:

@rabbit_router.subscriber(queue, exchange, retry=5)
async def handle_message(user: User, message: RabbitMessage):
    raise NackMessage()

Expected behavior
Message will be re-queued a maximum of retry times.

Observed behavior
Messaged is re-queued forever.

Additional context
Happens because of this logic in push_back_watcher.py, so the number of retries is never checked:

if not exc_type:
   await self.__ack()

elif isinstance(exc_val, SkipMessage):
    self.watcher.remove(self.message.message_id)

elif isinstance(exc_val, HandlerException):
    if isinstance(exc_val, AckMessage):
        await self.__ack()
    elif isinstance(exc_val, NackMessage):
        await self.__nack()
    elif isinstance(exc_val, RejectMessage):  # pragma: no branch
        await self.__reject()
    return True

elif self.watcher.is_max(self.message.message_id):
    await self.__reject()

Moreover, is message.nack() is called, exc_type is None and the first branch is taken calling await self.__ack()

[Lancetnik]

@sidjakinsboriss thank you for the Issue

I changed this behavior with 0.3.5 and now NackMessages also checks is_max condition

Btw, don't worry about default self.__ack - all messages can't be acked twice, so if you call ack/nack/reject manually, all following calls just do nothing