Seems to me that's exactly what a good scientific definition would be
No, quite the opposite. Good science synthesizes observations into coherent theory that can then be used to understand things that have not been observed before and make predictions beyond observation. Your definition does not do this.
Why not suppose a planet where water is HeO2? Diamonds are made from lead? we would still call the diamonds and water, right? Do you see any difference in "supposing an imaginary" observation, and real observations? I guess not.
HeO2 is not a possible compound. Diamond is a compound of Carbon and cannot be made from lead. This is basic chemistry. There is every reason, on the other hand, to suppose that the selection of DNA and ATP are arbitrary rather than necessary. On another planet they are unlikely to be repeated.
Your definition is so far removed from what we think of as life that it will produce absurd conclusions; the imaginary world of Delton-four is a simple illustration of where it is likely to break down. It's called a "thought experiment" and is a key part of science.