Only use split tree algorithm to reduce kernel code size.

[ROCm/rccl commit: 02929cffb6]

projects/rccl/src/collectives/device/all_reduce.h

@@ -691,8 +691,7 @@ struct RunWorkElement<ncclFuncAllReduce, T, RedOp, NCCL_ALGO_RING, NCCL_PROTO_LL
 template<typename T, typename RedOp>
 struct RunWorkElement<ncclFuncAllReduce, T, RedOp, NCCL_ALGO_TREE, NCCL_PROTO_LL> {
   __device__ __forceinline__ void run(ncclWorkElem *args) {
-    if (args->pad_0 == 0) runTreeUpDown<T, RedOp, ProtoLL>(args);
-    else runTreeSplit<T, RedOp, ProtoLL>(args);
+    runTreeSplit<T, RedOp, ProtoLL>(args);
   }
 };