ActiveSupport::Cache::RedisCacheStore should support Redis instances for `redis` kwarg

[as3richa]

Summary

Fixes a bug in #31134

ActiveSupport::Cache::RedisCacheStore should support operating upon a Redis instance given in its initializer (via the redis keyword argument). Ref: https://github.com/rails/rails/blob/master/activesupport/lib/active_support/cache/redis_cache_store.rb#L109-L116

In actual fact, because Redis.new.respond_to?(:call), RedisCacheStore will instead treat a Redis instance like a proc, calling it to populate its @redis: https://github.com/rails/rails/blob/master/activesupport/lib/active_support/cache/redis_cache_store.rb#L121-L122

Obviously this doesn't have the correct semantics. The test case added in my first commit fails with a Redis::TimeoutError (if indeed a Redis instance is running on 127.0.0.1:6379).

With the fix given in my second commit, RedisCacheStore#redis is precisely equal to the Redis instance given to the initializer, if indeed one was given.

Other Information

I'm a first-time contributor 😁 please let me know if I need to update any changelog or open this PR against a branch other than master.

[as3richa]

I left this branch untouched. My rationale is that someone may want to provide an object that is workalike with Redis (ie. implementing get, set, etc. with a different backend) but that doesn't inherit from Redis. Should this branch remain?

[matthewd]

Hmm... the triple branch does seem a bit ugly.

If we're going to have to use an is_a? check to distinguish them, maybe we'd be better changing the existing condition to is_a?(Proc)?

[as3richa]

Sure, SGTM

[as3richa]

Pushed

[jeremy]

Thanks @as3richa! Backported to 5-2-stable @ 2ebd68b