Fix trace conflict (#39421)

* add trace op

* bug fix

* bug fix; test=develop

* thrust bug fix; test=develop

* remove useless register; test=develop

* fix bug; test=develop

* update trace kernel; test=develop

* move kernel args to trace_sig; test=develop

* try to fix trace kernel conflict; test=develop

paddle/pten/kernels/gpu/trace_kernel.cu

@@ -33,9 +33,8 @@ void TraceKernel(const Context& ctx,
     auto stream = ctx.stream();
     std::vector<int> reduce_dims;
     reduce_dims.push_back(out->dims().size());
-    kernels::
-        TensorReduceFunctorImpl<T, T, kps::AddFunctor, kps::IdentityFunctor<T>>(
-            ctx, diag, out, kps::IdentityFunctor<T>(), reduce_dims, stream);
+    kernels::TensorReduceImpl<T, T, kps::AddFunctor, kps::IdentityFunctor<T>>(
+        ctx, diag, out, kps::IdentityFunctor<T>(), reduce_dims, stream);
   } else {
     paddle::operators::math::SetConstant<Context, T> functor;
     functor(ctx, out, static_cast<T>(0));