Some good points/questions being raised, and hopefully will help narrow down the problem.
In the meantime, I thought I'd add this note about the jet pump in case this isn't well known.
The diagram is from the 2006 service manual, but applies generally to Subaru and other makes that use a saddle-bag style fuel tank. (Although some makes use different transfer arrangements, including electric pumps).
In the diagram, we see that if the jet becomes blocked or for some related reason fuel isn't being transferred sufficiently from the left side to the right, the fuel pump, which is on the right, could eventually drain that side until its pick-up tube is no longer immersed in fuel.
There are fuel level sensors on both sides of the tank, and while the sensor on the right side would be at its lowest possible point when the engine quits, the sensor on the left side would still be registering the fuel left in that side. The two sensors are wired in series, so their respective "readings" are added. Empty on the right but 1/4 full on the left (or something like that) gives a 1/4 full reading on the gauge.
(You might be wondering why the LOW FUEL warning light does not come on when the right side is empty. On earlier models, the low fuel warning light was activated by a thermistor sensor on the fuel pump pick up tube. When the fuel level dropped below the sensor, it's temperature would change (it's in air rather than in liquid fuel) and this would turn the light on. Consequently, the light would/should come on if the right side of the tank is drained but not the left. However, on the later models, such as this 2006, the low fuel warning light is controlled by a microprocessor in the instrument panel that monitors the fuel gauge, and does not have any sensor in the tank itself. It turns the light on when the gauge gets down to a preset level. But because the gauge is still reading the remaining fuel in the left side of the tank, the light is not turned on.)
The difference being experienced between driving in the city (no problem), and driving on the highway ("runs out of gas"), could be related to the rate at which fuel is used by the engine.
The pump delivers fuel at a fairly high, steady, rate. In city driving, much of this is returned to the fuel tank via the jet pump, so a properly operating jet pump system would work efficiently and there will be a strong transfer of fuel from left to right.
However, in highway driving, especially at steady cruise, the difference between the pump fuel delivery rate and the engine consumption can be less, on average, than in the city driving. For example, in city driving the pump continues to work when the engine is idling and not consuming much fuel, so during this time the jet pump transfer would be at a high rate. But at steady highway driving, the amount of fuel being returned to the tank is less so, again on average, there could be somewhat less flow from left to right.
When the jet pump transfer system is working properly, this difference probably isn't of any significance. But if the jet pump or the flow of fuel transfer from left to right isn't as good as it should be, the difference during high speed highway driving could lead to the right side becoming drained before all the available fuel from the left side is transferred over. Once the pump can no longer draw up fuel through its pick-up tube, there's no way to correct the side-to-side fuel imbalance because the pump needs fuel to get the jet pump siphoning to work.
I recall that some years ago there was a case of fuel starvation that was traced to the jet pump. I don't think I've seen it confirmed again since, although there have been problem posts that could suggest this was the cause.
Again not saying this is the problem, but thought it could be among the possibilities.
