Fix C++ -Werror regressions in llama runner (#19326) (#19326)

Summary:

Fixes 3 `-Werror` diagnostics that broke the qualcomm llama runner build
on
`cfg:android-arm64-clang19-no-san` and disabled the following test infra
targets:

- `xplat/executorch/examples/qualcomm/oss_scripts/llama:runner_lib`
-
`xplat/executorch/examples/qualcomm/oss_scripts/llama:runner_lib_static`
-
`xplat/executorch/examples/qualcomm/oss_scripts/llama:qnn_llama_runner`
-
`xplat/executorch/examples/qualcomm/oss_scripts/llama:qnn_llama_runner_static`

Three diagnostics fixed:

1. `-Wreorder-ctor` in `runner.cpp`: `attention_sink_rope_module_` is
declared as the 2nd field of `Runner<T>` (right after `module_`) but the
constructor initializer list appended it last, after `tokenizer_`. Moved
it to the correct position in the init list to match declaration order.
   Recent regression introduced in the attention-sink diff (#16574).

2. `-Woverloaded-virtual` in `lhd_token_generator.h` and
   `multimodal_lhd_token_generator.h`: the derived classes define a
`prepare_io(std::vector<uint64_t>, std::vector<int32_t>)` overload that
   hides the base class virtual `prepare_io(uint64_t, int64_t)`. Added a
   `using TokenGenerator<T>::prepare_io;` (and equivalent for the
multimodal hierarchy) declaration so the base virtual stays in scope and
the warning is silenced without changing behavior. Latent bug surfaced
   by the clang19 toolchain bump.

3. `-Wdelete-non-abstract-non-virtual-dtor` in `prompt_processor.h`:
   `PromptProcessor<T>` has virtual member functions but no virtual
   destructor, so deleting via `std::unique_ptr<PromptProcessor<T>>` in
   `Runner` was undefined behavior under strict warnings. Added
   `virtual ~PromptProcessor() = default;` mirroring the pattern already
   used in `TokenGenerator` (`token_generator.h`). Also transitively
   fixes `MultimodalPromptProcessor<T>`.

Reviewed By: rascani

Differential Revision: D103991803

Author: psiddh

examples/qualcomm/oss_scripts/llama/runner/lhd_token_generator.h

@@ -102,6 +102,9 @@ class LhdTokenGenerator : public TokenGenerator<T> {
       AttentionSinkRopeRunner* attention_sink_rope_runner) override;
 
  private:
+  // Bring base class's virtual prepare_io into scope so the overload below
+  // does not hide it (-Woverloaded-virtual).
+  using TokenGenerator<T>::prepare_io;
   /**
    * @brief Fill in I/O buffers with prompt token and position.
    * @param cur_token Current token.

examples/qualcomm/oss_scripts/llama/runner/multimodal_runner/multimodal_lhd_token_generator.h

@@ -108,6 +108,9 @@ class MultimodalLhdTokenGenerator
       AttentionSinkRopeRunner* attention_sink_rope_runner) override;
 
  private:
+  // Bring base class's virtual prepare_io into scope so the overload below
+  // does not hide it (-Woverloaded-virtual).
+  using TokenGenerator<T>::prepare_io;
   /**
    * @brief Fill in I/O buffers with prompt token and position.
    * @param cur_token Current token.

examples/qualcomm/oss_scripts/llama/runner/prompt_processor.h

@@ -40,6 +40,8 @@ class PromptProcessor {
       const std::string& method_name,
       Metadata metadata);
 
+  virtual ~PromptProcessor() = default;
+
   /**
    * @brief Initialize I/O tensor and allocate I/O data buffer.
    * @param buffer_manager Pointer to IMemAlloc instance; by default, it uses a

examples/qualcomm/oss_scripts/llama/runner/runner.cpp

@@ -102,6 +102,7 @@ Runner<T>::Runner(
     std::unique_ptr<tokenizers::Tokenizer> tokenizer,
     std::unique_ptr<executorch::extension::Module> attention_sink_rope_module)
     : module_(std::move(module)),
+      attention_sink_rope_module_(std::move(attention_sink_rope_module)),
       ngram_(ngram),
       window_(window),
       gcap_(gcap),
@@ -111,8 +112,7 @@ Runner<T>::Runner(
       temperature_(temperature),
       eval_mode_(static_cast<EvalMode>(eval_mode)),
       shared_buffer_(shared_buffer),
-      tokenizer_(std::move(tokenizer)),
-      attention_sink_rope_module_(std::move(attention_sink_rope_module)) {
+      tokenizer_(std::move(tokenizer)) {
   stats_.reset();
 
   if (decoder_model_version == "llama2") {