add no_need_buffer into yaml for view op. (#73815)

Author: liuruyan

paddle/phi/ops/yaml/backward.yaml

@@ -3859,6 +3859,7 @@
   kernel :
     func : view_dtype_grad
     data_type : out_grad
+  no_need_buffer: input
 
 - backward_op : warpctc_grad
   forward : warpctc (Tensor logits, Tensor label, Tensor logits_length, Tensor labels_length, int blank = 0, bool norm_by_times = false) -> Tensor(loss), Tensor(warpctcgrad)

paddle/phi/ops/yaml/inconsistent/dygraph_backward.yaml

@@ -365,6 +365,7 @@
   args : (Tensor input, Tensor out_grad, int64_t[] dims = {})
   output : Tensor(input_grad)
   invoke: view_shape(out_grad, input.shape())
+  no_need_buffer: input
 
 - backward_op: fused_gemm_epilogue_grad
   forward : fused_gemm_epilogue(Tensor x, Tensor y, Tensor bias, bool trans_x, bool trans_y, str activation) -> Tensor(out), Tensor(reserve_space)