Update on "[ET-VK] Adding all tensor packing support for repeat op."

This diff updates Executorch Vulkan backend's `repeat` operation to support width, height and channel packed tensors.
It also updates the op_registry.py file to indicate `repeat` operation supports all packing.

Differential Revision: [D71477633](https://our.internmc.facebook.com/intern/diff/D71477633/)

[ghstack-poisoned]

Author: trivedivivek

backends/vulkan/runtime/graph/ops/glsl/copy_packed_dim_offset.glsl

@@ -45,7 +45,7 @@ const lowp int packed_dim = unhash_packed_dim(out_layout);
 ${layout_declare_spec_const(C, "int", "in_layout", "DEFAULT_LAYOUT")}
 const lowp ivec4 in_axis_map = unhash_axis_map(in_layout);
 
-${layout_declare_spec_const(C, "bool", "repeat", "false")}
+${layout_declare_spec_const(C, "int", "repeat", "0")}
 
 void no_repeat_copy(ivec3 pos) {
   // Position in input tensor
@@ -229,7 +229,7 @@ void main() {
     return;
   }
 
-  if (repeat) {
+  if (repeat == 1) {
     repeat_copy(pos);
   } else {
     no_repeat_copy(pos);

backends/vulkan/runtime/graph/ops/impl/Copy.cpp

@@ -151,7 +151,7 @@ void add_copy_packed_dim_offset_node(
       // Parameter buffers
       {},
       // Specialization Constants
-      {graph.hashed_layout_of(out), graph.hashed_layout_of(in), repeat},
+      {graph.hashed_layout_of(out), graph.hashed_layout_of(in), repeat ? 1 : 0},
       nullptr,
       {},
       {