Update on "[llm] Support different shape of input_pos"

For huggingface models, `forward()` is taking `tokens` as well as `cache_positions`, which is a list of cache indices. This is different than the .pte files `export_llama` gives, which are taking `tokens` and `input_pos` where `input_pos` is a scalar tensor.

This PR adds support inside `text_decoder_runner.cpp` to handle both shapes of `input_pos`/`cache_positions`.

To make the logic more generic without relying on extra metadata, here I'm adding the logic of inspecting method meta and input tensor info, to make a decision if we want to feed in `input_pos` or `cache_position`.

Differential Revision: [D77203700](https://our.internmc.facebook.com/intern/diff/D77203700/)

[ghstack-poisoned]

Author: larryliu0820

kernels/portable/cpu/util/arange_util.cpp

@@ -12,7 +12,7 @@ namespace torch::executor::native {
 #define ET_ARANGE_IMPL(ctx, start, numel, step, out, op_name)               \
   ET_SWITCH_REALHBF16_TYPES(out.scalar_type(), ctx, op_name, CTYPE, [&]() { \
     auto out_data = out.mutable_data_ptr<CTYPE>();                          \
-    for (size_t i = 0; i < numel; ++i) {                                    \
+    for (Tensor::SizesType i = 0; i < numel; ++i) {                         \
       out_data[i] = static_cast<CTYPE>(start + i * step);                   \
     }                                                                       \
   })