[MXNET-860] Remove std::moves that have no affect

.clang-tidy

@@ -55,7 +55,7 @@ Checks: >
 # In order to trigger an error, you must have a rule defined both in checks and in this section.
 WarningsAsErrors: >
     cppcoreguidelines-no-malloc, modernize-use-nullptr, performance-unnecessary-copy-initialization,
-    modernize-use-emplace
+    modernize-use-emplace, performance-move-const-arg
 
 # Todo: define a better regex match that includes most project headers, but excludes third party
 # code.

src/executor/infer_graph_attr_pass.cc

@@ -330,7 +330,7 @@ nnvm::Graph InferShape(nnvm::Graph&& graph,
     graph.attrs["shape_inputs"] = std::make_shared<any>(std::move(shape_inputs));
   }
   if (shape_attr_key.length() != 0) {
-    graph.attrs["shape_attr_key"] = std::make_shared<any>(std::move(shape_attr_key));
+    graph.attrs["shape_attr_key"] = std::make_shared<any>(shape_attr_key);
   }
   return InferAttr<nnvm::TShape, nnvm::FInferShape>(
       std::move(graph), nnvm::TShape(),
@@ -348,7 +348,7 @@ nnvm::Graph InferType(nnvm::Graph&& graph,
     graph.attrs["dtype_inputs"] = std::make_shared<any>(std::move(dtype_inputs));
   }
   if (dtype_attr_key.length() != 0) {
-    graph.attrs["dtype_attr_key"] = std::make_shared<any>(std::move(dtype_attr_key));
+    graph.attrs["dtype_attr_key"] = std::make_shared<any>(dtype_attr_key);
   }
   return InferAttr<int, nnvm::FInferType>(
       std::move(graph), -1,
@@ -366,7 +366,7 @@ nnvm::Graph InferStorageType(nnvm::Graph&& graph,
     graph.attrs["storage_type_inputs"] = std::make_shared<any>(std::move(storage_type_inputs));
   }
   if (storage_type_attr_key.length() != 0) {
-    graph.attrs["storage_type_attr_key"] = std::make_shared<any>(std::move(storage_type_attr_key));
+    graph.attrs["storage_type_attr_key"] = std::make_shared<any>(storage_type_attr_key);
   }
   // initialize unknown values for dispatch modes
   if (graph.attrs.count("dispatch_mode") == 0) {

src/operator/tensor/elemwise_unary_op.h

@@ -126,10 +126,10 @@ class OpBase {
     in_blobs.reserve(inputs.size());
     out_blobs.reserve(outputs.size());
     for (size_t i = 0, n = inputs.size(); i < n; ++i) {
-      in_blobs.emplace_back(std::move(inputs[i].data()));
+      in_blobs.emplace_back(inputs[i].data());
     }
     for (size_t i = 0, n = outputs.size(); i < n; ++i) {
-      out_blobs.emplace_back(std::move(outputs[i].data()));
+      out_blobs.emplace_back(outputs[i].data());
     }
     computer(attrs, ctx, in_blobs, req, out_blobs);
   }