Hard to diagnose, whether examining on-site or remotely from the description . . .
All the following may already be known; regrets if I'm telling something already obvious.
My understanding only; could be in error.
Basically, the carb vent line provides atmospheric pressure to the underside of the diaphragm, and, I think, to the surface of the fuel in the float bowl (so fuel can flow through the jets from venturi vacuum). Don't know if you have a Generation 1 or 2; Generation 1 carb vent line opening was about 10" above riding surface; left in stock configuration, subject to flame-out when water crossing level was deeper. Solution: Tee mod, providing alternate path to atmosphere when tube bottom was submerged.
Generation 2 carb vent opening is about 30" above riding surface; less likely to stop up from water crossings up to that depth. Tee mod not so vital for this model.
As to which way air flows in the carb vent tube, the answer is: BOTH ways. When venturi vacuum raises the slide, air flows IN; when venturi vacuum falls and the slide lowers, air flows OUT. Not a lot of flow, in either case; as mentioned, only enough to maintain atmospheric pressure on the underside of the diaphragm.
Given this scheme, an occluded (as in, submerged, stopped up, or pinched) carb vent hose will honk up your carb's ability to deliver mixture proportional to venturi vacuum.
Corrections and clarifications welcomed; regrets if I merely state the obvious and known.
Alternate cause of symptoms described: Stopped-up fuel tank cap vent; impedes fuel flow into carb by restricting air flow to replace fuel used from tank.
Another thing to check: Petcock fuel flow. Since the problem occurs at higher rpm, the petcock may not be delivering at the desired flow rate.
Regardless: You now have a CLEAN CARB!