Reality is constant slight of hand. Material circumstances are constantly in flux. Relationships are constantly in flux. You don't need conscious decision.
While self identification is naturally part of being trans, a much greater element of my concern is social identification without concern for that. We can conceive situations for androgynous individuals without any gender identity issues too, but these are more contrived and lead to a greater incidence of debating the example.
I would hold that if something appears to be a woman, it is a woman until those appearances no longer hold. This is of course, in reference to the mystified social category of womanhood rather than any given physical criteria. That is to say, the definition of a woman is constantly context-dependent and not always articulable. (Which is part of where the fun of delineating "actual womanhood" comes from, because if you say motherhood you have the infertile, if you say chromosomes you have outlying conditions, and so on and so fourth, because the reality is you're trying to articulate a "you just know" social category.)
i feel some analogy to the DSM is in order.
Until recently, you had all these neat little categories. You had aspergers, autism, PDD-NOS, etc. These were all distinct ways of categorising an underlying behavioural and medical reality, but they created all sorts of arbitrary borders because of social presumptions. (i.e. it was expected that aspergers meant higher functioning, when you could get cases just as bad as those considered autism, with the distinction arising because of how the descriptors were added together. in result, aspergers kids could be rejected for services they needed because those were "autism services" and even if the conditions were recognized as the "autism spectrum" the diagnosis was aspergers.)
then with the DSM-V these were all folded together into a single diagnosis of 'autism spectrum disorder', despite the underlying conditions not changing, despite the only material reality shift being a change on a page, the entire categorisation was altered. Yet despite this, the change filters down, because now rather than trying to split the difference between various similar conditions you can take consideration of the actual problems any given individual shows and provide services on that basis.
it is along a similar basis i would draw this entire thing together on. in part, we could say that this is a purely definitional, linguistic or categorisation argument - but as with the autism example, categorisations are important. (although because i'm eating, I'll avoid trying to contrive some examples of how fluid the autism definition is because it's far too easy to turn it into a series of jokes.)
in an ad-hoc addendum that will undoubtedly get disproportionate engagement if recognized: the problem with materialism is that it doesn't recognize most people aren't materialists and that this impacts the way they behave on a level that can't simply be buffed out of calculation. unless we want to push down to "thoughts are material" levels.)