Link to 'fire an input source' algorithm, explicitly construct frame by Manishearth · Pull Request #797 · immersive-web/webxr

It's a bit tedious to create the frame separately each time. I could also create the frame in the algorithm, however we may need to more strongly imply that the frame refers to poses from the time of the event, given that the algorithm is always called within a queued task (and thus the frame should not refer to poses from the time of the task)

Yeah, I can see that this is a bit of a pain, but I think your logic here is sound and that this captures the intended timing well.