I wouldn't say I deny the importance of biology at all. What I do deny is binary and simplistic thinking about that biology, because I think that is likely to hurt people who in one way or another don't fit the assumed norms of those categories. I obviously agree that assuming everyone can be approximated to a white European man is pretty awful: I don't see the argument for advocating fixing that by turning one pigeonhole into two, when we could do so much better by using our capacity to describe and model the complexity of people's realities and experiences. I don't think one can detach the sorts of concerns you raise from gender, too: the reason for poor outcomes for women medically is based on the wider oppression of women as a gender category.
I also don't think this is a case of hard cases making bad law: when it comes to issues of medical rights and identity, the law needs to cover everyone to create a reasonable measure of equality: at the very least, it needs a strong case that including the edge cases will produce actual harms to the majority, which in my view hasn't really been made here.
On your example: firstly, biomedical studies aren't in any sense a legal documentation issue, and it had been my understanding that was the focus of our discussion. Second, if you're doing biomedical studies, it is important to know a range of details about the subjects' biological sex characteristics, but I think those are things you should be asking them and if necessary verifying at the time of the study. That may be best done by using a heuristic of asking people what sex they were assigned at birth, but it may be best done by checking or asking about their chromosomes or which hormones they produce or take supplements of or their height and weight or their bone structure or their genital structure or any number of other potentially relevant features. At best, in biomedical terms, a binary sex marker is a rough heuristic that we have to use for the sake of time and cost: I don't think we should elevate it to the point where we treat it as a social class, because then we're just reinventing a new binary form of gender with all the problems that implies.
I am yet to see anyone explain how the position of "sex isn't simple, so get some actual data and do your best to cover the things you're actually studying" is a threat to accessible design.
Regarding ID models, the answer frankly is that nobody in a serious governing position seems to have thought about it in depth, and outside those circles really the only pitches on the table are "binary sex for everything, get rid of gender wokery" (from the right and the conservative-feminists), and "maybe switch to a trinary gender system across most forms of ID" (from some liberals who care about these issues) and "remove gender and sex from all forms of ID" (which I've seen from a few gender abolitionist left types, not to be confused with gender criticals who are in the binary sex only camp). Nobody so far as I'm aware has published any comprehensive reform scheme, or proposed any differentiation between different forms of ID, and the current differences in which ID requires or reflects what are due to circumstance not intention.