I think the reason it doesn't count as a measurement is that the the gravitational effects are so weak the no-far-off-particle and yes-far-off-particle outcomes overlap substantially.

You start with a cross-section of where a "measurer" particle might be, and if the far-off "to-be-measured" particle is present its gravity shifts the cross-section of the measurer particle's position very slightly, but there's still a lot of overlap with the original cross-section. A proper measurement would have nearly no overlap, instead of almost complete overlap.