Merge ivalue::Future's markCompleted and markCompletedWithDataPtrs (pytorch#56512)

Summary:
Pull Request resolved: pytorch#56512

I don't know if there was a reason to keep them separate, but since the former deferred to the latter, it seems to me that we can get the exact same behavior by merging them and making the `data_ptrs` argument optional (by giving it a default value).
ghstack-source-id: 127035767

Test Plan: Unit tests

Reviewed By: mrshenli

Differential Revision: D27861069

fbshipit-source-id: 93a49d6959b65a8d4ab9b31accce90bf30cd441e

Author: lw

aten/src/ATen/core/ivalue_inl.h

@@ -357,22 +357,16 @@ struct C10_EXPORT ivalue::Future : c10::intrusive_ptr_target {
   }
 
   /**
-   * Explicitly mark the future as completed with the output value.
-   */
-  void markCompleted(IValue value) {
-    markCompletedWithDataPtrs(std::move(value));
-  }
-
-  /**
-   * Explicitly mark the future as completed with the output value and DataPtrs.
-   * The data_ptrs contains storage pointers for all tensors in IValue, which
-   * will be passed to preMarkCompletedHook. Some subclass, like CUDAFuture,
-   * uses these DataPtrs to synchronize CUDA streams. You only need to provide
-   * data_ptrs when 1) DataPtrs cannot be extracted through
-   * IValue::getSubValues() or 2) customized DataPtrs extraction is more
+   * Explicitly mark the future as completed with the output value. Optionally,
+   * the storage pointers for all tensors in IValue can be passed as well, and
+   * it will be passed to preMarkCompletedHook. Some subclass, like CUDAFuture,
+   * uses these DataPtrs to synchronize CUDA streams. If data_ptrs isn't given
+   * the subclass will attempt to extract it from the value, if it needs. Thus
+   * one only needs to provide data_ptrs when 1) DataPtrs cannot be extracted
+   * through IValue::getSubValues() or 2) customized DataPtrs extraction is more
    * efficient.
    */
-  void markCompletedWithDataPtrs(
+  void markCompleted(
       IValue value,
       c10::optional<std::vector<std::reference_wrapper<const at::DataPtr>>>
           data_ptrs = c10::nullopt) {

torch/csrc/distributed/rpc/python_functions.cpp

@@ -164,7 +164,7 @@ c10::intrusive_ptr<JitFuture> toPyJitFuture(
               return;
             }
 
-            child->markCompletedWithDataPtrs(ivalue, std::move(dataPtrs));
+            child->markCompleted(ivalue, std::move(dataPtrs));
           }
         }));
     return child;