[miguel-heygen]

The runtime test job ran cli, core and engine tests one after another and took 7 to 9.5 minutes against a 10 minute limit, so some PRs were cancelled right after their tests passed.

The cli tests now run in a parallel Test (cli) job; Test (runtime) keeps core and engine. The required Test check waits on both and fails if either does not succeed. No required check names change.

Before and after job times are in this PR's own CI run.

[github-actions]

Edit accuracy: accurate 271 (base branch 271), smooth 151 of those

The gate passes.
Smoothness is reported in the artifact, not gated. A case fails only if it fails 2 of 3 runs.

Quarantined, measured but not gated (0)

[heygengenesis]

Verdict: REQUEST CHANGES · 2 blocker · 0 important · 0 nits · head 562c323e

The shard split itself is right. Test now waits on test-cli and fails closed, and in this run Test (cli) took 5m15s and Test (runtime) 4m39s. There are two blockers, both inline: the required Format check is red, and the new job does not install the numpy that #6327 needs.

Scope

Read the full diff and grepped for other test-runtime / Test (runtime) consumers; Test is the only aggregator. Reproduced the oxfmt 0.41.0 output locally. git merge-tree of this head with #6327 d6ee1cfc merges cleanly, so git will not flag the numpy break.

[heygengenesis]

Verdict: APPROVE — 2 resolved, 0 open, 0 new · 562c323e → a378cafc. CI is green: Test (runtime) took 4m35s and Test (cli) 4m52s.