The finishing times of non-binary runners will probably not show a bimodal distribution. Marathon times for mixed-sex groups generally don't: the mode for the group may be somewhere between the modes for male and female runners, but those single-sex modes won't be apparent unless the data are separated by sex in the first place.
Interestingly (though not particularly relevant to sex and gender), marathon times can show multimodal distribution unrelated to sex (or any other characteristic of the runners). Modes are typically found at about 3, 3:30, 4, 4:30 and 5, because many runners try so hard to achieve one of these 'milestone' times. You need large numbers of runners to notice this sort of detail though, and non-binary marathon runners are relatively few.
https://marathonhandbook.com/average-marathon-time/