Fix text_llm_runner kv cache pos count and use it for generate()

[pytorchbot]

Summary

pos_ should advance by prefill and generated prompt size.

Test plan

CI

cc @larryliu0820 @mergennachin @cccclai @helunwencser @jackzhxng

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15295

• 📄 Preview Python docs built from this PR

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

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• AWS was down, GHA infrastructure effected / recovering

❌ 1 New Failure, 2 Cancelled Jobs

As of commit 2bcc2cb with merge base e0dda90 ():

NEW FAILURE - The following job has failed:

• pull / test-eval_llama-wikitext-linux / linux-job (gh)
  RuntimeError: Command docker exec -t d9afc8e4a1aaef5756e5048b251a7cde266e7ee10b0a58760da38e56572903c7 /exec failed with exit code 4

CANCELLED JOBS - The following jobs were cancelled. Please retry:

• pull / unittest / macos / macos-job (gh)
  ##[error]The operation was canceled.
• pull / unittest-editable / macos / macos-job (gh)
  ##[error]The operation was canceled.

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

[jackzhxng]

Will this be in a patch?

[kirklandsign]

Will this be in a patch?

Yes