A few weeks ago I created a discussion titled "How does Bell's inequalities rule out realism."
Essentially my question was pertaining to how does removing realism retain locality and not violet Bell's inequality.

Someone answered with the this,

I'm not really happy with this explanation, but I understand it. However as I've been thinking it doesn't seem like this would produce the same predictable results that we observe in entanglement. I noticed that someone brought it up as well, saying,

"If in an experiment locality is given. And realism is rejected in accounting for Bell violations, then how is it that this non realism is so consistent and derivable?"

Without realism, physics becomes the physics of observation - what you see, not what is there. Without going too deep into the subject, you would say that the rules of the universe are such that the observer always see things as consistent with QM - and since this does not need to reflect anything "real" in the world, locality is never an issue. However, if you do go beyond this shallow analysis, you need to ask whether you are remembering things correctly. When you attempt to find a correlation between A and B, you need to rely on the record you made of the first one - and later, when examining the second one, how would you know that what you see of that record is real?

Please keep in mind that "realism" in this case has a special meaning, different than the ordinary usage. Further, there are a number of ways in which non-realism can be pictured or expressed, each being a violation of "realism".

They mostly share the idea that the act of measurement (whatever that is) is a fundamental part of the quantum mechanical context. Therefore quantum elements do not exist independently of how they are observed.

The idea that "the observer shapes reality elsewhere" is in tune with many themes in QM. But we really have no idea how that occurs (mechanically). Perhaps there is a splitting of worlds. Perhaps there is retrocausality.

I have question from first quote above in original post: "Suppose two measurements at A and B are correlated at spacelike separation"
"One observer cannot see both measurement results at once as long as A and B are spacelike separated"
In the above diagram at time point when measurements are completed at A and B,
light signals with measurement outcomes could be sent to observer at source.

The original statement appeared to imply that the reason disregarding realism retains locality and does not violate bell's inequality is because the observer would not be able to view the spin of both particles at once. However it seems at a conceptual level, and as the illustration indicates, that this is not the case. The observer would be able to see the measurement results at the same time.
Any explanation or is the interpretation wrong?

The point here is subtle and surprisingly difficult to state clearly. The correlation can only appear in the future light cone of both measurements and therefore does not necessarily imply non-locality. We make measurement A - no locality problem. Someone else makes measurement B - no locality problem, unless you assert that one measurement affects the result of the other. Now we can bring these measurements together using ordinary non-local methods (carry the lab notebooks to a desk in the middle of the lab, radio the results to the other experimenter or a third party, ....) and the correlation is observed. Thus, the observation of the correlation requires no non-locality at all - one observer, at one point in spacetime, was handed two measurement results that got to him through ordinary channels with no faster than light travel.

Realism is the assumption that the measurement outcomes existed before the correlation was observed. If you want this assumption, you'll find non-locality forced on you.

Nugatory already answered it, but maybe this is simpler and clearer: seeing both results at once is not just seeing both at the same time. It means to see both instantly when they occur. And for one observer that is only possible if they occur at the same place.

But of course two observers can do both measurements at the same time (arguably), or (non arguably) at least, at times that no communication could have taken place, due to the limited speed of light. As those observers can compare their results afterwards, non-realism is a rather extreme explanation (are they both hallucinating?).

First off removing (the assumption of) realism does not necessarily retain locality.
Let me make some simple definitions to see we are on the same page.
Locality: No influence can be transmitted faster than light speed.
Realism: Bob flips a fair coin, and in the case he flips heads, will go thru some process to obtain a value Bh = either +1 or -1. Similarly, if he flipped tails he may go thru a different process to obtain Bt = +1 or -1. Before he flips he may not know what Bh and Bt will turn out to be, but they will be be something and he can consider the term Bh + Bt, for example.
Bell's Inequality: Alice, who is separated from Bob (see post #4) does something similar to get value Ah and At also = +1 or -1.
Theorem: Assuming realism and locality, the probability that for an arbitrary pair of flips by Alice and Bob that Ah•Bh = Ah•Bt = At•Bh = 1, and At•Bt = -1 will occur is ≤ 3/4. (Bell's Inequality)

Real world QM experiments violate Bell's Inequality. No physical experiment or physical theory can disprove a mathematical theorem. Thus the assumption of both realism and locality (the hypothesis of the theorem) must not jibe with experiment. So one or the other or both don't jibe. Without the assumption of both the theorem can't be proved, in which case Bell's Inequality has no valid logical status. If we assume locality is valid (no one knows if that is the case in spite of relativity theory) then of course realism doesn't jibe with experiment.