[llm] Add a generic text only LLM runner

Pull Request resolved: #11342

Introducing `text_llm_runner`. This can be used to run all text only decoder only LLM models supported by ExecuTorch.

* Metadata is being read out from the .pte file and being used to construct the runner object.
* examples/models/llama/runner.h[.cpp] only contains a simple wrapper around `text_llm_runner.h[.cpp]`.


In next PRs I will move examples/models/phi-3-mini/runner to use the generic runner.

Will look into QNN and MediaTek runners as well.
ghstack-source-id: 288346995
@exported-using-ghexport

Differential Revision: [D75910889](https://our.internmc.facebook.com/intern/diff/D75910889/)

Author: larryliu0820

examples/demo-apps/apple_ios/LLaMA/LLaMARunner/LLaMARunner/Exported/LLaMARunner.mm

@@ -14,6 +14,7 @@
 
 using executorch::extension::llm::GenerationConfig;
 using executorch::extension::llm::Image;
+using executorch::extension::llm::TextLLMRunner;
 using executorch::runtime::Error;
 
 NSErrorDomain const LLaMARunnerErrorDomain = @"LLaMARunnerErrorDomain";
@@ -23,15 +24,15 @@ @interface LLaMARunner ()<ExecuTorchLogSink>
 @end
 
 @implementation LLaMARunner {
-  std::unique_ptr<example::Runner> _runner;
+  std::unique_ptr<TextLLMRunner> _runner;
 }
 
 - (instancetype)initWithModelPath:(NSString*)modelPath
                     tokenizerPath:(NSString*)tokenizerPath {
   self = [super init];
   if (self) {
     [ExecuTorchLog.sharedLog addSink:self];
-    _runner = example::Runner::create(
+    _runner = example::create_llama_runner(
         modelPath.UTF8String, tokenizerPath.UTF8String);
   }
   return self;

examples/models/llama/main.cpp

@@ -81,8 +81,13 @@ int32_t main(int32_t argc, char** argv) {
   }
 #endif
   // create llama runner
-  std::unique_ptr<example::Runner> runner =
-      example::Runner::create(model_path, tokenizer_path, data_path);
+  std::unique_ptr<::executorch::extension::llm::TextLLMRunner> runner =
+      example::create_llama_runner(model_path, tokenizer_path, data_path);
+
+  if (runner == nullptr) {
+    ET_LOG(Error, "Failed to create llama runner");
+    return 1;
+  }
 
   if (warmup) {
     runner->warmup(prompt, /*max_new_tokens=*/seq_len);