[OR] Bug: Strange snap behavior in profile

Using Geopak SS4 .878.

Given: an existing terrain model, set as active terrain. (This will work with a jagged profile like one derived from a surface--probably any complex profile. It is not triggered if your profile is a simple parabola.)

1. Construct a horizontal civil element. For demonstration purposes, make it something like 300 units long or longer and ensure that it isn't running too far out of the terrain model, if at all.
2. Open this civil element's profile model.
3. Profile Offset Transition. Use the active terrain surface. Use any offset value (I'm using +3). Have it begin at station 200 and have it run for 100 units.
4. Profile Line Between Points. Snap to each end of the offset created in step 3. Intuitively, you know that selecting this element should reveal that it begins at 200 and ends at 300 with a length of 100.
5. Select the offset element created in step 3, and move the begin point some distance toward the beginning of the alignment. It doesn't matter so much exactly where, but it needs to be far enough that it will have reached a break line in the terrain model. In my case, moving it 25 feet was enough.
6. Now select the straight line created in step 4. Observe that this straight line is no longer following the begin point on the offset element. The begin station will match the station where you'll find a break line in the terrain model. In my case, 199.2157. The end station shifted with the offset element's end station.
7. Now if I repeat step 5 again and move it another 25 feet, I see that the straight line element continues to begin at 199.2157. The end station shifts with the offset element's end station.
8. Now if I repeat step 5 except now move it 50 feet forward instead of back, I see that the straight line element's begin point moves to another break point in the active terrain model. This time, it is at 296.5232. The end station shifts with the offset element's end station, now back at 300. That means the element is only 3.4768 feet long (longitudinally).
9. If I inch the offset element forward to 301 then 302, etc., the end of the offset advances as well, and with it the end of the straight line advances. When the end of the offset element gets past the next break line, the beginning of the straight line hits that next break line as well.

This is where I lose the trail. I'm not even sure sure what it's picking up at this point. This behavior is very confusing.

So now, from what I can tell, I can't trust profile snaps.