What's varying most significantly here, as the n2yo path clearly illustrates, is the latitude (North and South of the Equator), not the longitude (East and West on the equator).
Correct, as they are only doing East-West station-keeping, which probably doesn't require much fuel to make the minor adjustments necessary.
Just a little clarification here, are you asking why the pattern has changed so much from, as an example, the pattern on the left, to what it is now, on the right? NOTE: the image on the left is an old one of SES3, the one on the right, F2 currently. Also note the there are not equal in scale as to the size of the orbit variations.
If so, SES3 was at the time "in the box" so any adjustments East-West and North-South would be carried out when necessary. F2, of course, is only getting East-West adjustments.
Looking at the elliptical orbit of SES3, any adjustments I would assume (and it is an assumption only) would be done East-West when the satellite was in the middle of the orbit (or could be they are done at the top-bottom of the orbit!). I would also assume that any North-South adjustments would be done at the top or bottom limit of the elliptical orbit. These station-keeping measures would shrink the elliptical orbit. I have also seen some satellite orbits that are almost circular, which would suggest that station-keeping in both directions had been done fairly recently.
With F2's orbit, I would imagine that any East-West adjustments could be done at pretty much any time necessary. I have also seen other inclined satellite orbits which are a squashed (East-West) figure eight (obviously they are also getting East-West adjustments only).