I should also add that sometimes I don't really want to zoom the timeline down either.
When I'm analysing these things, I usually have 2 or 3 frames in use, where the other frames keep track of changes to INP or resyncs, or ES's, CRC's, re-transmission counters, etc - so I can see the interplay between multiple graphs at the same time. Having correlated timelines makes this easy ... and getting matching zoomed timelines ends up quite awkward - especially if you have to fiddle to move a spike off one side.