Why doesn't Faster Than Sound information transfer imply time travel or violate causality? I am struggling to understand the answer to What are some scenarios where FTL information transfer would violate causality?
In particular, I have not seen a satisfactory rebuttal to the comment from user @Malvineous, which I find intuitively very compelling:   

But a reference frame existing where the events appear out of order
  doesn't seem to violate causality, it just means that there is a delay
  between the events happening and that observer seeing them, and the
  delays are such that the events look like they happened out of order,
  even though they didn't. [...]

So, I concocted a scenario that tries to replicate the thought experiment using sound waves, instead of light.
Imagine a sleeping dog, a supersonic airplane travelling towards the dog, and a blind human observer a few sound-seconds away. Further assume that the blind observer has a very keen sense of hearing - in both intensity and direction. The airplane is flying at some negligible distance above ground. Disregard the physiological effects on the dog (no dogs were harmed in this thought experiment). Assume the dog's barking sound is super-caninely loud (so it can be heard miles away).
In Experiment 1, the observer's path to the dog is at a 90-degree angle to the flight path (i.e. consider a "T" shape setup where the top of the T is a runway, the observer stands at the bottom tip of the T, and the dog is in the middle of the runway). The supersonic airplane flies towards the dog, waking the dog when it flies over. The dog immediately wakes up and barks at the plane. The plane activates a dog alert siren and continues to fly away from the dog. The blind observer hears (after some delay): airplane flying towards dog, dog barking at plane, plane flying away from dog with dog-alert sirens blaring. The dog hears a BOOM, barks, then hears what will seem like two planes, simultaneously approaching and flying away, one on each side.
In Experiment 2, the observer moves closer to the end of the runway. The supersonic airplane flies towards the dog, waking the dog when it flies over. The dog immediately wakes up and barks at the plane. The plane activates a dog alert siren and continues to fly away from the dog. The blind observer hears (after some delay): plane flying away from dog with dog-alert sirens blaring, dog barking, airplane flying towards dog.
In Experiment 3, a sight-enabled observer is brought in, to describe events to the blind observer, as they are happening. There is now much-faster-than-sound information transfer. Experiment 2 is repeated, but now the sighted observer quickly tells the blind observer: "In 10 seconds, you are going to hear an airplane flying towards a sleeping dog". A moment later, "oh by the way, 2 seconds before that (7 seconds from now), you are going to hear the dog barking at the plane."
The dog, and each observers, each hear a different sequence of sounds. What differentiates this observed reordering of events, from causality violations that occur with hypothetical FTL travel in relativity?

To summarize that, why is it that when an observer sees events arrive in a different order than they occurred, it is considered a violation of causality in relativity, but not in slower-than-light physics?
 A: You are confusing the perception of the observer (and their limitations) with the actual speed with which information is transferred.  Just because a person or machine does not detect something does not mean it did not happen.
You are also allowing an observer (the sighted assistant) to make predictions.  Strictly speaking the sighted person cannot say that a sound will be heard in the future.  It's a prediction they made based on a model of what they see.
If you allow observers to make predictions (which is not the same as information being transferred) then you could also break the limit on no faster than light information transfer, because the prediction (which in reality could prove incorrect) would be (according to you !) information transferred FTL.
But that's not what physicists mean by FTL information transfer.  No predictions are allowed.  You have to wait for the actual information to arrive.
A: None of your scenarios are related to causality. In order to get a signal-based causality problem you need to send a signal to a receiver, and then have the receiver reply such that the reply arrives before the signal was sent. 
So in your examples a causality violation would be something like the blind man hearing the dog bark and then shouting to the airplane to throw him a bone before he barks. 
However, the underlying question that you seem to be hinting at is a bit different. You correctly assert that you can send signals by sound or by light. So I believe that your underlying question is something along the lines of: if faster than light signals violate causality then why don’t faster than sound signals do that too? The reason is that the speed of light is the same in all inertial reference frames, whereas the speed of sound is not.
UPDATE: responding to your update, it is not true that “when an observer sees events arrive in a different order than they occurred, it is considered a violation of causality in relativity”. For instance, a star went supernova centuries before I was born and my 30th birthday party was after, but I saw the party before I saw the supernova. There is no causality violation there. 
