Arm backend: Improve while support

[Erik-Lundell]

• Support additional_args for while by normalizing them to carried_args in a pass.
• Support not using all while outputs.

cc @freddan80 @per @zingo @oscarandersson8218 @digantdesai

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/16287

• 📄 Preview Python docs built from this PR

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

❌ 2 New Failures, 1 Unrelated Failure

As of commit 9479cb5 with merge base d263897 ():

NEW FAILURES - The following jobs have failed:

• Build Presets / apple (macos) / build (gh)
• pull / test-openvino-linux / linux-job (gh)
  RuntimeError: Command docker exec -t fec453fd9ed3944617a584a41c17c258ea78203a1a25d50d11c494aa747171e6 /exec failed with exit code 1

UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:

• pull / android / run-emulator (gh) (#16137)
  Timeout waiting for emulator to boot.

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

[zingo]

Oh no :( But mean abs error: is really low so it seem like it's flakey and unrelated, could be worth to just retrigger when the other tests have runned all the way.

FAILED backends/arm/test/ops/test_sdpa.py::test_sdpa_vgf_quant - AssertionError: Output 0 does not match reference output.
Given atol: 0.006777485202997923, rtol: 0.001.
Output tensor shape: torch.Size([1, 3, 197, 64]), dtype: torch.float32
Difference: max: 0.01155497133731842, abs: 0.01155497133731842, mean abs error: 5.8043679665079164e-06.
-- Model vs. Reference --
Numel: 37824, 37824
Median: 0.0, 0.0
Mean: 0.0018774077386055387, 0.0018777132326251095
Max: 0.7048531770706177, 0.7048531770706177
Min: -0.6932982206344604, -0.6932982206344604

[Erik-Lundell]

Since this pr touches while and the failing test i spda, I would assume it's a flaky test. Let me retrigger

[zingo]

Worked, other fails unrelated.