remove unused tensors from VK model's graph

Summary:
We implemented [operators fusion](#3769) (`conv+bn`) which fused `conv` and `bn`'s weights and biases, but the old parameters are not deleted. Hence we saw that VK model's size is nearly twice large as CPU's.

As regards mobilenet_v2, before this diff CPU vs VK: 14M vs 22M. After this diff, both of them have 14M.

Differential Revision: D60257047

backends/vulkan/serialization/vulkan_graph_builder.py

@@ -262,12 +262,14 @@ def get_or_create_value_for(self, arg: _Argument):
             raise RuntimeError(f"Cannot create value for arg of type {type(arg)}")
 
     def process_placeholder_node(self, node: Node) -> None:
-        ids = self.create_node_value(node)
-        if not self.is_param_node(node):
-            if isinstance(ids, int):
-                self.input_ids.append(ids)
-            else:
-                self.input_ids += ids
+        # ignores any tensors that don't get used in any ops
+        if len(node.users) != 0:
+            ids = self.create_node_value(node)
+            if not self.is_param_node(node):
+                if isinstance(ids, int):
+                    self.input_ids.append(ids)
+                else:
+                    self.input_ids += ids
 
     def process_getitem_node(self, node: Node) -> None:
         # Find ValueList id from the collection node.