Not to forget the 'both' option!
Okay, lame joke. But still, an at least half-serious question: Might there not be folks who, at times, in certain contexts, identify as male, and at other times, other contexts, as female? With neither times, neither set of contexts, overriding the other in frequency or importance to them? And, indeed, in some times and some contexts identifying as neither?
Come to that, I suggest that, perhaps, we all answer to this model, in a way. Even the most staid and prosaic of us will surely, for example, identify as male in certain contexts, and as sex-is-entirely-irrelevant in others, right, at least these two even if never female? Just a question of degree, I guess, when it comes to which particular aspect -- if any! -- stands out in individual cases.
That could be one way of looking at it. Of course, in cases where at most times and most places what one answers to is 'male' -- for example -- well, that is then what one 'is'.
So sure, it isn't binary, at least not in principle. (Albeit it may often be that, binary I mean, in practice. Or, if not 'often', then 'sometimes'; who knows, perhaps even 'rarely'. That much at least, the answer to that last question, can easily, given sufficient resources, be ascertained pretty much objectively. Provided, of course, the question itself is generally found to be a reasonable one.)
.
Okay, one way to delink this from the physical might be via VR. Immerse a guy in a SL-like VR environment for long enough, where they can be who or what they choose, and see how they actually, in practice, act. What they identify as, there. Given a long enough duration of this, as an experiment, we might be able to find out how frequent, or otherwise, non-binary identification is?