It doesn't make any difference who accelerated away from who initially, it's only acceleration after the two clocks depart that determines which one will be behind when they meet again. The perspective of a frame where the earth is initially moving at 0.99c, and the clock that accelerates away from it is actually slowing down
initially, is every bit as valid as the perspective of the earth's frame.
No, the way you are using "reference state", basically to mean using the initial conditions to determine which frame's answer is more "correct" than the others, is not an idea that any physicists would find useful, and you haven't given any justification for why
the initial conditions should cause us to prefer one frame over another. It's totally arbitrary, as far as I can see (and note my comments earlier about the choice of which state to label the 'initial' one as being equally arbitrary).
I don't think pervect understood what you meant by "reference state", since he seemed to think it just meant the same thing as "reference frame", when actually you are using it to mean an initial state which you use to determine a preferred frame. I'm pretty sure he wouldn't agree with this notion of yours that the initial state tells us that one frame's perspective is more valid than others.