Update on "Add update_quantized_cache op"

Why?
- ton of copies due to functionalization
- mutable buffer support without such custom inplace ops will results in giant
  copies at the end
- Making inplace ops work will likely take longer and not clear safe path

Differential Revision: [D62301838](https://our.internmc.facebook.com/intern/diff/D62301838/)

[ghstack-poisoned]

Author: kimishpatel

kernels/quantized/cpu/op_choose_qparams.cpp

@@ -73,12 +73,12 @@ void check_quantize_per_tensor_args(
     }
     ET_CHECK_MSG(
         scale_out.size(input.dim() - 1) == 1,
-        "Exepcted scale to have only one element at dimentions %ld but got %zd",
+        "Exepcted scale to have only one element at dimentions %zd but got %zd",
         input.dim() - 1,
         scale_out.size(input.dim() - 1));
     ET_CHECK_MSG(
         zero_point_out.size(input.dim() - 1) == 1,
-        "Exepcted zero point to have only one element at dimentions %ld but got %zd",
+        "Exepcted zero point to have only one element at dimentions %zd but got %zd",
         input.dim() - 1,
         zero_point_out.size(input.dim() - 1));
   } else {