Move trunk.yml, periodic.yml and nightly.yml to linux_job_v3

[huydhn]

Fourth of six splitting up #22107. Stacked on #22247.

Also picks up the last linux_job.yml (v1) call site, in trunk's test-arm-backend-zephyr.

The two arm backend jobs that raise fs.inotify.max_user_watches now tolerate failure, since that is a node-level setting an unprivileged pod cannot change.

Authored with Claude Code.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/22248

• 📄 Preview Python docs built from this PR

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

✅ You can merge normally! (1 Unrelated Failure)

As of commit 79bf15b with merge base c8d5189 ():

BROKEN TRUNK - The following job failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / test-samsung-models-linux / linux-job (gh) (trunk failure)
  AttributeError: type object 'Test_Milestone_MobileBertFinetune' has no attribute 'model_cache_dir'

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

[huydhn]

Last commit temporarily points trunk.yml (11 refs), periodic.yml and nightly.yml at pytorch/test-infra#8683 (osdc/hf-cache-writable). Reverts to @main before landing.

Same caveats as #22247: a pull_request run already sees the label in the event payload, so this exercises the refresh path rather than the ciflow-tag resolution, and it will not seed the shared cache until meta-pytorch/pytorch-gha-infra#1463 lands.

trunk.yml is the interesting one here — it carries a ciflow/trunk tag trigger, so if a ciflow/trunk/22248 tag gets pushed, that run does exercise the tag resolution, which is the path with no coverage in test-infra.