Try something simple...
If you can take the cover off the skimer while the pump is in operation, watch how the water is being taken into the line.
With the cross-area differential, you might be creating a vortex in the strainer body; thus, allowing the pump to suck air when the return is open all of the way.
I know that this will happen in the smaller waterbaths we have in the lab under certain conditions: I have a waterbath where the drain is on the bottom... it has 1/4" opening. The entire piping used to be 1/4". The pump burnt out, we changed it out with a new pump from the manufacture. The new pump was rated the same, just used 1/2" fittings.... so we used a 1/4" to 1/2" adaptor between the drain tubing and new pump. We upsized the remaining fittings and tubing (the return hole actually was a 1/4" fitting with gasket.. the 1/2" worked without modification). We were actually trying to reduce the force that the return had as it pushed the glass beakers around the waterbath which cause other issues!
The pump would hiccup... it loose prime, then prime, loose prime, prime... Strangest thing I'd ever seen... my lab manger at the time had never seen anything like it either and he'd been in labs 20 ot 30 years at that time. (always a warm fuzzy when the old timers scratch their heads... confirms that I'm not totally crazy... well, maybe)
What was happening was that the pump would pull a vortex, suck air, loose prime, then the water would back fill, and the whole thing cycle.
The manufacture had already solved the same problem for other customers. There were two solutions... the first was that we put a vortex breaker in the waterbath
Vortex breaker). The second was that we replaced part of the return tubing with the 1/4".
The vortex breaker alone worked just fine; however, we weren't getting enough mixing in the bath; thus the second fix - what I found though was that I could take that vortex breaker out of the bottom drain after we reduced the return size. The return to the 1/4" tubing was also suggested by the manufacture; however, when explained why we went to the 1/2" opening, that is when they sent the vortex breaker.
-wc