Superadmin-only: push a human-graded audit run into the LangSmith eval dataset — one example per
question, landing in the incoming split (plus the run's cohort split) for review-then-promote
to golden in the LangSmith UI. Gated behind the global enableEvalRuns flag, like
evals.runExperiment.
Superadmin-only: push a human-graded audit run into the LangSmith eval dataset — one example per question, landing in the
incomingsplit (plus the run's cohort split) for review-then-promote togoldenin the LangSmith UI. Gated behind the globalenableEvalRunsflag, likeevals.runExperiment.