Mathematics are indeed arbitrary in the sense that the axioms can be chosen however you wish. HOWEVER, once a set of axioms is chosen, all the logical consequences of it follow and it is not possible to then wave your hands and say "it's just a theoretical construct and I can believe whatever I like" towards results you personally disagree with.
IN your case, you are making a basic mistake. If S is a bounded sequence that can be contained in an open set X, then the sup and inf of S lie NOT in X, but in the closure of X.
Proof: (I'll just show it for the supremum since the proof is the same for the other extrema.)
Since S is a bounded sequence, sup S is not infinite. By definition if s is an element of S, s is less than or equal to sup S.
If s = sup S, this is trivial because sup S is in X, and thus, is in the closure of X.
If s < sup S, S is an infinite sequence. More importantly, by the definition of supremum, for any r > 0, we can always find an n such that s_n is the nth element of S and sup S - s_n < r. However, since the above is true for any r, sup S by definition exists in the closure of X.
Now back to your problem. The closure of the set X [0.9, 1) is [0.9, 1]. Thus, if S = (0.9, 0.99, 0.999, ....), then S is in X and sup S = lim n->infinity of s_n is in the set [0.9, 1]. Since clearly sup S is not in X, sup S must be in the set cl(X) - X = {1}. Hence, 0.999.... = 1.