The problem is that you don't have enough pressure in your panels at the lower flow rate since the 2-story return to the pool is negative pressure that needs to be overcome. Otherwise, the valve at the top of the panels opens up and lets air in. One way of handling this is to restrict the flow at the bottom of the return from the panels so that the pressure will be higher. You can have the same flow rate, but the restriction will increase the pressure. It's similar to what happens when a pump is connected to narrower piping or to something with a flow restrictor. To get the same flow rate, the pressure must be higher.
If your panels even when hot could handle negative pressure without collapsing, then a different sort of valve that only opened to let air in when the pressure exceeded a certain negative pressure (negative with respect to air; i.e. lower than air pressure) could be used, but that is not normally done (I write later on about moving the air intake to a lower height instead of at the top of the panes). Instead, flow restriction is added to increase pressure in the panels. Ideally, such flow restriction would be dynamic so that at higher flow rates (pump RPM) less restriction would be used (I think there are spring valve restrictors that do this, but others know about such things better than I).
If each story is 12 feet and your panels are on a roof that is another 6 feet high at the top, then that's 30 feet which is 13 PSI. The formula I use for my Intelliflo relating Head (feet of water) to RPM and GPM is the following:
Head (feet of water) = (RPM/350)2 - GPM2/470
With 2500 RPM and 50 GPM I get 45.7 feet of head which is obviously more than 30 and likely higher than the height at the top of your panels where the air release valve is likely placed. You didn't say the RPM when you are at 50 GPM, but if I assume 1500 RPM at 28 GPM then I get a Feet of Head of only 16.7 which obviously is not enough.
From an energy efficiency point of view for the pump and from an efficiency point of view for the solar panels, not adding flow restrictions and instead just having a higher flow rate is better. Adding flow restrictions needlessly wastes pump energy through the friction/turbulence from the restrictor. There is probably a compromise you can make where you could lower the point of the air intake for draining of the panels since it may not need to be quite at the top of the panels since some amount of negative pressure there is probably OK. Then you could somewhat lower your pump speed and flow rate for more energy savings.
Mark (mas985) is the expert in this area and hopefully he'll come along and give you some advice.