Recognize the lock state of multiple semaphores using the same key

[eugenk]

When you have many semaphore instances with the same key, only the one that called lock knows that the semaphore is locked, but all of them should find the correct state with locked?. Example of the faulty behavior:

[1] pry(main)> redis_sema = Redis::Semaphore.new(:my_sema)
=> #<Redis::Semaphore:0x007fc6d8c8aae0
 @expiration=nil,
 @name=:my_sema,
 @redis=#<Redis client v3.3.0 for redis://127.0.0.1:6379/0>,
 @resource_count=1,
 @stale_client_timeout=nil,
 @tokens=[],
 @use_local_time=nil>
[2] pry(main)> redis_sema.lock
=> "0"
[3] pry(main)> redis_sema2 = Redis::Semaphore.new(:my_sema)
=> #<Redis::Semaphore:0x007fc6d8b480d8
 @expiration=nil,
 @name=:my_sema2,
 @redis=#<Redis client v3.3.0 for redis://127.0.0.1:6379/0>,
 @resource_count=1,
 @stale_client_timeout=nil,
 @tokens=[],
 @use_local_time=nil>
[4] pry(main)> redis_sema2.locked?
=> false
[5] pry(main)> redis_sema.locked?
=> true

[NickDiMucci]

I'm actually running into the same issue. It would have been great if your PR was merged in.