Use larger memory CPU runners for build jobs

[jrhemstad]

Description

Updates non-GPU CI jobs to use the larger memory CPU runner labels (cpu16m) instead of the standard cpu16 labels to avoid OOMs during builds.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[jrhemstad]

/ok to test 7e6f374

[jrhemstad]

/ok to test 424668e

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: b13a1672-e5f5-45f8-97dd-9ee8a734ffa7

📥 Commits

Reviewing files that changed from the base of the PR and between 84c2c01 and 424668e.

📒 Files selected for processing (1)

• .github/actions/workflow-build/build-workflow.py

---

📝 Walkthrough

Summary by CodeRabbit

• Bug Fixes
  • Updated CPU-only CI dispatch jobs to use the correct runner label, improving job assignment and build reliability.

Walkthrough

important: verify cpu16m runner label exists in CI runner pool before merging; a mismatched label will cause dispatch jobs to fail to schedule.

Changes the CPU-only runner label in generate_dispatch_job_runner from a cpu16 suffix to cpu16m, affecting runner assignment for non-GPU CI dispatch jobs.

Changes

Runner Label Update

Layer / File(s)	Summary
CPU runner suffix change .github/actions/workflow-build/build-workflow.py	Runner label string changed from *-cpu16 to *-cpu16m for CPU-only job dispatch.

Related PRs: None identified.

Suggested labels: ci, github-actions

Suggested reviewers: None identified.

Poem
A rabbit hops through YAML lines,
Changes one small suffix, "m" it finds.
Sixteen cores now marked anew,
cpu16m, a label true.

---

_{Comment @coderabbitai help to get the list of available commands.}

[trxcllnt]

Other alternative is to create a swapfile (PR here). I imagine cpu16m runners are more costly and availability may be limited.

[github-actions]

😬 CI Workflow Results

🟥 Finished in 2h 50m: Pass: 99%/506 | Total: 16d 15h | Max: 2h 35m | Hits: 54%/1797913

See results here.

[Jacobfaib]

Other alternative is to create a swapfile (#9739). I imagine cpu16 runners are more costly and availability may be limited.

If that also unblocks CI (which based on your 1-shot green CI it does) while keeping existing runner capacity, I'd be in favor of this.

@jrhemstad what do you think?