使能Iterate异步接口避免AIC/AIV同步依赖

【优先级】：高

【描述】在AIC和AIV混合编程中，调用Matmul Iterate或者IterateAll时，AIV发送消息到AIC启动Matmul计算。若Iterator<sync=true>同步方式，如图1，每次调用都会触发一次消息发送，而通过Iterator<sync=false>异步方式，仅第一次需要发送消息，后续无需发送消息，从而减少Cube与Vector核间交互，减少核间通信开销。

因此，mix场景推荐使用Iterator<false>或者IterateAll<false>异步接口。

【反例】

TQueBind<TPosition::VECOUT, TPosition::A1>  qCubeInA1;
TQueBind<TPosition::CO2, TPosition::VECIN>  qVecIn;

auto a1InUB = qCubeInA1.AllocTensor<half>();
mm.SetTensorA(a1InUB);
qCubeInA1.FreeTensor(a1InUB);

while( qCubeInA1.template Iterator()){
    auto cInUB = qVecIn.AllocTensor<float>();
    qCubeInA1.GetTensorC(cInUB);
    qVecIn.EnQue(cInUB);
    cInUB = qVecIn.Deque<float>();
    vector_op(cInUB);
    ...
    qVecIn.FreeTensor(cInUB)
}

【正例】

TQueBind<TPosition::VECOUT, TPosition::A1>  qCubeInA1;
TQueBind<TPosition::CO2, TPosition::VECIN>  qVecIn;

auto a1InUB = qCubeInA1.AllocTensor<half>();
mm.SetTensorA(a1InUB);
qCubeInA1.FreeTensor(a1InUB);

while( qCubeInA1.template Iterator<false>() ){
    auto cInUB = qVecIn.AllocTensor<float>();
    qCubeInA1.GetTensorC(cInUB);
    qVecIn.EnQue(cInUB);
    cInUB = qVecIn.Deque<float>();
    vector_op(cInUB);
    ...
    qVecIn.FreeTensor(cInUB)
}