Adding new ExecuTorch Executor that supports HTP-delegated models (#2322)

Summary:
Pull Request resolved: https://github.com/pytorch/executorch/pull/2322

# Summary

As part of our efforts to migrate Supernova Media Enhancement models to ExecuTorch, we ran into an issue with the existing Executor, which does not support using two buffers to pass data between the CPU and HTP for HTP-delegated models.

# This diff
This diff adds another Executor that uses [HexagonDelegatedModel](https://www.internalfb.com/code/fbsource/[bcf2921bfadf004815ad89d363d510f7bb8cc202]/arvr/projects/bolt/bolt/nn/executorch/Utils.h) util to allocate memory and load methods for HTP-delegated models.

We are trying to land and cherry pick this into the V4 branch cut, so we are adding this into the same target as ExecutorchExecutor to avoid issues right now.

# Future plans
After the V4 cherry pick deadline, we can go ahead and lift the executor code out of the pyspeech codebase and into another library to avoid tech debt.

bypass-github-export-checks
bypass-github-pytorch-ci-checks
bypass-github-executorch-ci-checks

Reviewed By: tarun292, doggeral

Differential Revision: D54658355

fbshipit-source-id: c159ac8775d7485f66fb99fd8c698573899dc8a9