True, interesting idea.

[Long edit]

Thinking further about this idea, I realize this may even mitigate the catch 22 problem of very low orbits (<180km): the lower the orbit, the larger the drag and the required thrust power, meaning the solar arrays must be bigger, which in turn further increases the drag... Calculations suggests that with current solar array and thruster technology, flying lower than 150km with this concept is impossible.

But with an elliptic orbit, energy from the solar arrays can be stored on the low-drag portion of the orbit too and used during the perigee dip, thus decreasing the requirements in terms of solar arrays area.

This would have the added benefit of increased efficiency due to the Oberth effect; however, I'm still not sure you could use it for unassisted interplanetary flight. The last orbit, by definition, must occur before the craft passes Earth escape velocity -- the question is, in that last pass through perigee, can you get enough delta V to make it to another planet? Otherwise, you'd need supplemental propellant. It's still useful, it's just not something I would describe as "revolutionary" for interplanetary travel.

Since the TWR of electric thrusters tends to be pretty abysmal, my gut is that you probably couldn't scale up the thruster well enough to bounce between planets without that supplemental propellant.

That being said, as others have mentioned, this would be really quite interesting for stationkeeping at low orbital altitudes, particularly for small satellites.