Cortex_m backend: Add script for building test runner

[AdrianLundell]

cc @digantdesai @freddan80 @per @zingo @oscarandersson8218

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/14750

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 New Failure

As of commit f9c12e6 with merge base 3b16bc1 ():

NEW FAILURE - The following job has failed:

• pull / test-samsung-models-linux / linux-job (gh)
  RuntimeError: Command docker exec -t 7bb6054de3fb15f0cdd1d6ff086c4caf08a677e379eda9cf509f05bee84273d7 /exec failed with exit code 1

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[AdrianLundell]

Fails unrelated

[AdrianLundell]

@digantdesai @psiddh Even though this calls scripts in the arm directory I think it makes sense to add now, partly since it is more user friendly to use scripts in the same folder as you are working in and partly since it gives allows for more separation from the arm backend. For example we can add selected aten ops to the executor runner only when building it for the cortex_m backend easily.

[digantdesai]

100%. From user point of view, this is an independent backend and different from backends/arm. However, from implementation point of view, we do what makes most sense which is do rely on infrastructure from backends/{arm,xnnpack,transforms} etc. when it makes sense.