Record what the frozen workers bought the campaign

CHANGELOG.md

@@ -194,7 +194,7 @@ recorded trace hash moves, once, for every workload.
   the syncs Raft owes before it answers an RPC; drop the sync before an
   append is acknowledged and the first seed loses a committed entry.
 - Campaign evidence at scale, regenerable via `benchmarks/campaign.py`:
-  100,000 seeds of jobqueue chaos green in just over six minutes on a laptop,
+  100,000 seeds of jobqueue chaos green in just over four minutes on a laptop,
   every ablation caught with its failure density recorded, and 20 sampled
   failing seeds replaying with identical trace hashes across 100 re-runs
   apiece. A small nightly CI sweep keeps the numbers honest.

README.md

@@ -277,7 +277,7 @@ recording included — about 3.7× faster than the stock loop, which pays a
 selector syscall per iteration), compresses sleep-heavy workloads ~2,000×
 against wall clock, and with `--simloop-jobs` fanning seeds across
 processes, the full jobqueue chaos scenario sweeps 100,000 seeds in just
-over six minutes (~260 seeds/second) on an M4 MacBook Air. Methodology, numbers,
+over four minutes (~390 seeds/second) on an M4 MacBook Air. Methodology, numbers,
 and the campaign results:
 [benchmarks/README.md](https://github.com/dhruvl/simloop/blob/main/benchmarks/README.md).
 

benchmarks/README.md

@@ -119,9 +119,16 @@ Results, recorded 2026-08-04 on the M4 MacBook Air (10 jobs):
 
 | campaign | scale | result |
 |---|---|---|
-| green | 100,000 seeds, 6.3 min, 263.1 seeds/s | green — no invariant violated |
-| ablations | 6 mutations × 10,000 seeds, 2.7 min | every ablation caught, densities below |
-| replay stability | 20 failing seeds × 100 re-runs, 14.2 s | identical trace hash on every run |
+| green | 100,000 seeds, 4.2 min, 394.5 seeds/s | green — no invariant violated |
+| ablations | 6 mutations × 10,000 seeds, 3.0 min | every ablation caught, densities below |
+| replay stability | 20 failing seeds × 100 re-runs, 15.7 s | identical trace hash on every run |
+
+The green row is where the workers' frozen heap shows up: nearly every green
+seed reaches the settle phase and pays the end-of-run collections in full,
+so taking the imported heap out of them moved this number from 263 to 394
+seeds/s in one sitting on this machine. The ablation runs barely moved —
+most of those seeds fail within a few steps and never accumulate the work
+the collection was re-walking.
 
 Per-ablation failure density: