Sheep Guts
Small correction to your list of VMCA demonstration conditions:
FAR 25.149(c)(4) states that the "maximum sea level takeoff weight (or any lesser weight necessary to show VMC)" is to be used.
It's a pretty convoluted wording, because they're mxing up two things in the same section. But in practice it's the part in brackets that matters, because VMCA is generally higher at lighter weights, due primarily to the limited 5 degree bank angle and the lower weight reducing the sideslip which can be generated, this sideslip being helpful in trimming out the yawing moment from the engine asymmetry.
I recall seeing an Airbus paper which argued that higher weights should be considered, rather than the traditional demonstration at minimum practical weight, as the higher angles of attack at heavier weights could reduce rudder effectiveness in some cases. But I believe it's still generally held that light weight is more critical.
John is correct in stating that the dependency with flap is aircraft dependent; we do generally determine the dependency in the course of VMC analysis, but the AFM is often simplified to a single value agreed with the cert authorities.
FWIW, on three pretty similarly configured aircraft my company has recently tested, the relationships were:
1. No stated flap dependency, VMCG < VMCA
2. VMCA higher with larger flap angles. VMCG a single value between the two VMCA values
3. VMCA lower with increasing flap angle. VMCG higher with increasing flap angle. Values 'cross over' between flap settings