docs(design): close RJN-4 as N/A (RJN-3 was rename-only)

[oferchen]

Summary

• Adds docs/design/rjn-4-closure-2026-05-21.md as a per-tracker closure note for RJN-4 (chore: remove unnecessary #[allow(dead_code)] in deferred_sync.rs #2560).
• Decision: N/A. RJN-2 (PR refactor(engine): rename apply_chunk_parallel to apply_one_chunk for clarity (RJN-2) #4660, merged) was a pure rename of apply_chunk_parallel to apply_one_chunk; with zero semantic change there is no before/after cores-vs-throughput shape for RJN-4 to bench.
• Splits the RJN-4 half cleanly out of the combined RJN-3/4 closure (PR docs(design): defer RJN-3 (fanout) and RJN-4 (bench) as N/A after RJN-2 rename #4676) so the tracker has its own artifact; references RJN-1 (PR docs(audits): apply_chunk_parallel call sites and per-chunk dispatch benefit (RJN-1) #4656), RJN-2 (PR refactor(engine): rename apply_chunk_parallel to apply_one_chunk for clarity (RJN-2) #4660), PIP-3+5 (PR perf(transfer): enable parallel receive-delta by default via Path B heuristic (PIP-3 + PIP-5) #4666), and the sibling ABW-2/3/4 closure (PR docs(design): defer ABW-2/3/4 pending BR-3j.f bench evidence (ABW-1 audit closure) #4673) which carries the production-weighted scheduler-shape bench gate via BR-3j.f (feat(daemon): add max verbosity parsing and enforcement #2508).
• Re-open trigger documented: batched fan-out lands at the receiver and the PIP-3 / PIP-5 default-on path starts consuming a multi-chunk batch interface through apply_one_chunk. Until then, scheduler-shape bench effort belongs at apply_batch_parallel (BR-3i.f / BR-3j.f against parallel_receive_delta_perf), not the per-chunk entry point.

Test plan

• Docs-only change; cargo fmt --all is a no-op.
• Cross-references to predecessor PRs (docs(audits): apply_chunk_parallel call sites and per-chunk dispatch benefit (RJN-1) #4656, refactor(engine): rename apply_chunk_parallel to apply_one_chunk for clarity (RJN-2) #4660, perf(transfer): enable parallel receive-delta by default via Path B heuristic (PIP-3 + PIP-5) #4666, docs(design): defer ABW-2/3/4 pending BR-3j.f bench evidence (ABW-1 audit closure) #4673, docs(design): defer RJN-3 (fanout) and RJN-4 (bench) as N/A after RJN-2 rename #4676) and trackers (feat(daemon): add max verbosity parsing and enforcement #2508, chore: remove unnecessary #[allow(dead_code)] in deferred_sync.rs #2560) verified against master.
• Matches the writing style and structure of the existing deferred-N/A closure docs (abw-2-pipelined-verify-write-deferred-2026-05-21.md, rjn-3-4-fanout-deferred-2026-05-21.md, ffb-3-4-pip-2-closure-2026-05-21.md).