Migrate sampler to extension/llm (#4460)

Summary:
Pull Request resolved: #4460

Move sampler code to extension/llm so that it can be reused by
llava runner.

Reviewed By: helunwencser

Differential Revision: D60458803

fbshipit-source-id: ef8c4d7d3fed4f0777e5ba9cd403da8320efef5a

Author: larryliu0820

examples/models/llama2/runner/runner.h

@@ -17,7 +17,7 @@
 #include <string>
 #include <unordered_map>
 
-#include <executorch/examples/models/llama2/sampler/sampler.h>
+#include <executorch/extension/llm/sampler/sampler.h>
 #include <executorch/extension/llm/tokenizer/tokenizer.h>
 #include <executorch/extension/module/module.h>
 #include <executorch/extension/runner_util/managed_tensor.h>

examples/models/llama2/runner/targets.bzl

@@ -33,7 +33,7 @@ def define_common_targets():
             ],
             exported_deps = [
                 "//executorch/backends/xnnpack:xnnpack_backend",
-                "//executorch/examples/models/llama2/sampler:sampler" + aten_suffix,
+                "//executorch/extension/llm/sampler:sampler" + aten_suffix,
                 "//executorch/extension/evalue_util:print_evalue" + aten_suffix,
                 "//executorch/extension/runner_util:managed_tensor" + aten_suffix,
                 "//executorch/extension/module:module" + aten_suffix,

examples/models/llama2/sampler/TARGETS renamed to extension/llm/sampler/TARGETS

@@ -32,7 +32,7 @@
  * SOFTWARE.
  */
 
-#include <executorch/examples/models/llama2/sampler/sampler.h>
+#include <executorch/extension/llm/sampler/sampler.h>
 
 namespace torch {
 namespace executor {