Update on "[ET-VK] AOT logic for quantized conv2d"

### Changes

As title; this diff adds the necessary export logic required to enable fusing quantized convolution patterns into the custom ops introduced in the below diff.

Differential Revision: [D83437826](https://our.internmc.facebook.com/intern/diff/D83437826/)

[ghstack-poisoned]

Author: ssjia

backends/vulkan/test/custom_ops/CMakeLists.txt

@@ -96,4 +96,6 @@ if(TARGET vulkan_backend)
   add_operator_prototype(q4gsw_linear)
   add_operator_prototype(choose_qparams_per_row)
   add_operator_prototype(qdq8ta_conv2d_activations)
+  add_operator_prototype(q8ta_q8csw_q8to_conv2d)
+  add_operator_prototype(q8ta_q8csw_q8to_conv2d_dw)
 endif()