Fix for queue once lock key not being deleted properly

[kuba-lilz]

Problems: redis lacks flush

When running QueueOnce subclass task, I noticed the lock key never got deleted, event though QueueOnce.after_return was called and issued self.once_backend.clear_lock(key).
After examining it turned out that I had to add self.redis.flushall() to clear_lock(key) for key to be cleared properly.

Here are my libs versions:

In [1]: import redis, celery, celery_once

In [2]: redis.__version__
Out[2]: '3.2.0'

In [3]: celery.__version__
Out[3]: '4.2.1'

In [4]: celery_once.__version__
Out[4]: '2.1.0'

In this pull request I added redis.flushall() after lock delete request, which solves the issue.

[cameronmaske]

Hi @kuba-lilz flushall deletes all keys of all databases, I don't think it is the solution here.

This sounds like it could be related to #93 (the keys aren't generating correctly).
As I mentioned there, could you include the code snippet of the task and how it is called?

[kuba-lilz]

A bad implementation that tries to solve a problem that didn't exist, see discussion in #93