Okay here are two theories:
1)
- Why the drag decreases with wing sweep in subsonic speeds below M2.1:
The part of the wing behind the mach cone 'knows' the air is coming, so the airflow can adjust itself to the wingshape. This means no shock drag, but induced drag.
The smaller the sweep angle, the bigger part of the wing is outside the mach cone, which means shock drag on the part outside of the wing.
So a low swept wing at subsonic speeds: a lot of induced drag + a lot of shock drag = more drag in total.
- Why the drag increases with wing sweep in subsonic speeds above M2.1:
The faster the aircraft, the smaller the mach cone.
With a lower swept wing, more of the wing is now outside the cone. Now the drag is mainly shock drag again, and less induced drag: total drag is less.
2)
The higher the speed, the higher the compressibility effects, the higher the temperature, the higher the local speed of sound, the lower the local mach number.
So:
- Why the drag decreases with wing sweep in subsonic speeds below M2.1:
with a highly swept wing, less compressibility, local mach nr. doesn't decrease that much. With a straight wing: a lot of compressibility, local mach nr. decreases back to the transonic range, a lot of shock drag.
- Why the drag increases with wing sweep in subsonic speeds above M2.1:
At speeds higher than M2.1, the compressibility effects become greater, even at the highly swept wing, reducing the local mach nr. to the transonic range, meaning lots of shock drag.
The straight wing has even greater compressibility effects, reducing the local mach nr. to a number below the transonic range, meaning subsonic speed, no shock drag at all anymore.
Or maybe it's a combination of both theories.
Those theories are probably not correct, but I'd rather have a semi correct theory that at least explains it a bit, than no theory at all and just accept it.
But of course, counterarguments are very much welcome!