Update on "[ET-VK] Implement missing Vulkan operators for Parakeet TDT model"

Add missing operators needed for Parakeet TDT model support:

- New symint ops: sym_sub, sym_floordiv, sym_mul in SymIntOps.cpp;
  register operator.floordiv and operator.mul as ephemeral ops in
  op_registry.py
- New tensor ops: bitwise_not (via unary_op shader with uint8 DTYPE),
  logical_and (alias for bitwise_and dispatch)
- Improve _to_copy: expand dtype support to FP_INT_BOOL_T and use
  pick_io_storage_fn to restrict to CONTIGUOUS_BUFFER for non-fp
  conversions
- Fix where resize: compute output shape via broadcast across all tensor
  inputs instead of always using the second input's shape
- Add symint support to split: use extract_int_or_symint_list instead of
  get_int_list in resize_split_node and split_with_sizes_copy_default
- Mark scalar_tensor as supporting resize

Differential Revision: [D95970159](https://our.internmc.facebook.com/intern/diff/D95970159/)

cc manuelcandales digantdesai cbilgin

[ghstack-poisoned]

Author: ssjia

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

@@ -45,9 +45,9 @@ void main() {
 
   VEC4_T out_texel = VEC4_T(0);
 
-  int limit = min(
+  const int limit = min(
       4, out_meta.sizes[packed_dim] - out_tidx.data[packed_dim]);
-  for (int comp = 0; comp < 4; comp++) {
+  for (int comp = 0; comp < limit; comp++) {
     TensorIndex4D in_tidx = out_tidx;
     in_tidx.data = ivec4(
         out_tidx.data.x % in_meta.sizes.x,

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

@@ -28,6 +28,12 @@ void resize_expand_node(
   const std::vector<int64_t> target_sizes =
       graph->extract_int_or_symint_list(size_ref);
 
+  VK_CHECK_COND(
+      target_sizes.size() >= in_sizes.size(),
+      "expand: target sizes must have at least as many dims as input");
+  VK_CHECK_COND(
+      !target_sizes.empty(), "expand: target sizes must not be empty");
+
   const size_t dim_offset = target_sizes.size() - in_sizes.size();
   std::vector<int64_t> out_sizes(target_sizes.size());
   for (size_t i = 0; i < target_sizes.size(); i++) {