Improved storage capabilities

[Kariton]

because the current implementation of storage.read() or storage.save() do not work well when they run simultaneously i moved the _redis from storage class to redis.
now the entire redis class is available within the storage class.
furthermore storage.delete() is now a thing.

i have a command that fetches content from a website. this takes time.
to make sure a user never receives the same post twice i used to:

1. storage.read()
2. do stuff
3. storage_result.append(new_id)
4. storage.save()

if the user runs the same command twice in a row it might happen that both coroutines read the same data and only the last one is actually saved.

now this is possible:

if c.bot.storage.redis is not None: # 'None' is default if storage is not redis
    c.bot.storage.redis.smembers(storage_key) # use redis.Redis().smembers() - list the set of stored data
    c.bot.storage.redis.sadd(storage_key, item["id"]) # use redis.Redis().sadd() - add item to the end of the existing data

p.s.: sorry to spam your repo. i messed up my git settings... new pc and stuff...

[Era-Dorta]

Would you still like to merge this? I'm happy to review if so.

[Kariton]

Hey,
yes, i heavily depend on that concurrent coroutines and need the additional functionality.

for now i depend on my own fork and think others would, if they want to, profit from the extended storage capabilities as well.

[Era-Dorta]

Amazing, thanks! Let me set myself as reviewer and add a few comments :)

[Era-Dorta]

How about making this class abstract to ensure the methods are implemented?

Suggested change

	class Storage:
	class Storage(ABC):

[Era-Dorta]

I don't understand why the base Storage class needs to have the redis attribute defined. If anything, for consistency with the InMemoryStorage, it should be renamed to _storage

[Era-Dorta]

Totally optional but how do you feel about adding some unit testing for the new function?