Improve cutensor contraction for certain cases

cuTensor contraction performance may heavily depend on the contraction kernel rank. It is recommended to manually set [`CUTENSOR_PLAN_PREFERENCE_KERNEL_RANK`](https://docs.nvidia.com/cuda/cutensor/latest/api/types.html#_CPPv4N33cutensorPlanPreferenceAttribute_t36CUTENSOR_PLAN_PREFERENCE_KERNEL_RANKE) to 0 for optimal performance