Evaluate confirmed-commit completion classification overhead if count coverage changes #3332

Draft #3187 classifies local completion rejections so an explicitly confirmed child cannot abort another local commit. The default unconfirmed path retains Promise.all. The component measurement on Node 26.2.0 reported 1.014 → 1.274 µs for two already-resolved local completions (+0.260 µs), excluding storage and peer acknowledgement.

Evaluate whether production count coverage or a smaller aggregation permits removing that classification while preserving the tested child ordering and prompt local failures. Record the measured tradeoff if retaining it. The current implementation deliberately keeps the shared helper used by both engines; this is the accepted performance follow-up from #3187.

Review-Deferral: 428d4c0261177c16 severity=minor category=performance

_{🤖 Generated by OpenAI Codex; posted via @kriszyp.}