[STF] Low level interface for the cuda_kernel(_chain) construct

cudax/include/cuda/experimental/__stf/internal/context.cuh

@@ -127,6 +127,15 @@ class context
       return *this;
     }
 
+    template <typename... Args>
+    auto& add_kernel_desc(Args&&... args)
+    {
+      payload->*[&](auto& self) {
+        self.add_kernel_desc(::std::forward<Args>(args)...);
+      };
+      return *this;
+    }
+
     template <typename T>
     decltype(auto) get(size_t submitted_index) const
     {
@@ -135,6 +144,22 @@ class context
       };
     }
 
+    auto& start()
+    {
+      payload->*[&](auto& self) {
+        self.start();
+      };
+      return *this;
+    }
+
+    auto& end()
+    {
+      payload->*[&](auto& self) {
+        self.end();
+      };
+      return *this;
+    }
+
   private:
     ::std::variant<T1, T2> payload;
   };