refactor: extract ModeHLL files (no contract change)

[Code0987]

Design

docs/design/2026-09-02-refactor-hll.md

What landed

Same HLLAdd / HLLCount contract. File layout only:

File	Role
pkg/store/hll.go	store methods (out of memory.go)
pkg/engine/hll.go	public verbs
pkg/engine/hll_apply.go	owner write + inbox apply
pkg/engine/hll_cluster.go	forward / snapshot / GetOrLoad

hllx.Merge is now unexported (merge). Engine never called it.

Tests

go test ./pkg/hllx ./pkg/store ./pkg/engine

Local bench

BenchmarkStoreGetHit 2 allocs/op, BenchmarkEngineGetHit 15 allocs/op (flat).

Bench risk

No new Get/Peek logic. Shared smoke/micro ±10%. Get-hit allocs must stay flat.

[github-actions]

SuperCache bench vs main

Both sides ran once on this same GitHub runner (main, then this PR). Not a merge gate.

Baseline: 48e5f71 (2026-09-02T13:44:09Z) → this run: 48e5f71 (2026-09-02T13:42:58Z)

scbench smoke

cell	ops/s (main → PR)	Δ ops/s	p50	p99
get/hit n=1 c=1 uniform	7259 → 7274	+0.2%	137µs → 138µs	220µs → 217µs
get/hit n=1 c=10 uniform	28325 → 29123	+2.8%	327µs → 318µs	891µs → 869µs
get/hit n=3 c=10 uniform	33546 → 34478	+2.8%	263µs → 258µs	1.022ms → 930µs
miss/miss-cacheonly n=1 c=1 uniform	7041 → 7196	+2.2%	142µs → 143µs	242µs → 221µs
miss/miss-cacheonly n=1 c=10 uniform	28658 → 30328	+5.8%	323µs → 304µs	928µs → 887µs
set/put n=1 c=1 uniform	6887 → 6973	+1.3%	142µs → 142µs	257µs → 226µs
set/put n=1 c=10 uniform	27471 → 28188	+2.6%	336µs → 327µs	946µs → 922µs

Δ ops/s: green-ish is higher throughput. Latency: lower is better.

go test -bench (ns/op)

benchmark	ns/op (main → PR)	Δ	allocs/op
BenchmarkEngineBitCount	162 → 162	+0.1%	0 → 0
BenchmarkEngineBitGetHit	155 → 157	+0.9%	0 → 0
BenchmarkEngineBitGetHitParallel	194 → 187	-3.6%	0 → 0
BenchmarkEngineBitSet	483 → 484	+0.3%	2 → 2
BenchmarkEngineBitSetParallel	523 → 555	+6.2%	2 → 2
BenchmarkEngineCMSIncr	9189 → 9509	+3.5%	1 → 1
BenchmarkEngineCMSIncrN	11376 → 9218	-19.0%	1 → 1
BenchmarkEngineCMSIncrParallel	14030 → 13484	-3.9%	1 → 1
BenchmarkEngineCMSQueryHit	196 → 194	-0.9%	1 → 1
BenchmarkEngineCMSQueryHitParallel	240 → 238	-0.7%	1 → 1
BenchmarkEngineCounterGet	151 → 151	+0.1%	0 → 0
BenchmarkEngineCounterGetParallel	170 → 172	+1.3%	0 → 0
BenchmarkEngineDelete	819 → 818	-0.2%	6 → 6
BenchmarkEngineGeoAdd	12142 → 12511	+3.0%	8 → 8
BenchmarkEngineGeoCard	152 → 152	-0.1%	0 → 0
BenchmarkEngineGeoPosHit	97 → 94	-2.6%	0 → 0
BenchmarkEngineGeoRadius	334996 → 362684	+8.3%	1016 → 1016
BenchmarkEngineGetHit	1265 → 1243	-1.7%	15 → 15
BenchmarkEngineGetHitParallel	1088 → 1080	-0.7%	15 → 15
BenchmarkEngineGetMissCacheOnly	943 → 923	-2.0%	14 → 14
BenchmarkEngineGetMissCacheOnlyParallel	674 → 712	+5.7%	14 → 14
BenchmarkEngineGetMissLoadThrough	1683 → 1680	-0.2%	24 → 24
BenchmarkEngineGetMissLoadThroughParallel	1902 → 1986	+4.4%	23 → 23
BenchmarkEngineHGetAll	16412 → 15963	-2.7%	134 → 134
BenchmarkEngineHGetHit	107 → 105	-1.9%	1 → 1
BenchmarkEngineHGetHitParallel	136 → 132	-2.9%	1 → 1
BenchmarkEngineHGetMiss	164 → 167	+1.6%	0 → 0
BenchmarkEngineHLLAdd	2054 → 2042	-0.6%	1 → 1
BenchmarkEngineHLLAddParallel	3414 → 3306	-3.2%	1 → 1
BenchmarkEngineHLLCountHit	107704 → 107925	+0.2%	0 → 0
BenchmarkEngineHLLCountHitParallel	117117 → 120667	+3.0%	0 → 0
BenchmarkEngineHLen	151 → 159	+5.8%	0 → 0
BenchmarkEngineHSet	13307 → 13298	-0.1%	6 → 6
BenchmarkEngineHSetParallel	15284 → 13855	-9.3%	6 → 6
BenchmarkEngineIncr	366 → 363	-0.9%	1 → 1
BenchmarkEngineIncrParallel	430 → 403	-6.2%	1 → 1
BenchmarkEngineJsonGet	189 → 190	+0.7%	1 → 1
BenchmarkEngineJsonGetHit	466 → 464	-0.5%	5 → 5
BenchmarkEngineJsonGetHitParallel	550 → 558	+1.5%	5 → 5
BenchmarkEngineJsonSet	28846 → 28523	-1.1%	215 → 215
BenchmarkEngineJsonSetParallel	16883 → 17374	+2.9%	119 → 118
BenchmarkEngineLIndexHit	179 → 173	-3.4%	1 → 1
BenchmarkEngineLLen	156 → 151	-3.1%	0 → 0
BenchmarkEngineLPush	10802 → 10524	-2.6%	18 → 18
BenchmarkEngineLRange	479 → 479	-0.1%	11 → 11
BenchmarkEnginePut	1585 → 1560	-1.6%	18 → 18
BenchmarkEnginePutParallel	1386 → 1441	+4.0%	18 → 18
BenchmarkEngineSetAdd	11847 → 11914	+0.6%	4 → 4
BenchmarkEngineSetCard	202 → 156	-22.5%	0 → 0
BenchmarkEngineSetContainsHit	98 → 98	+0.1%	0 → 0
BenchmarkEngineSetContainsHitParallel	117 → 120	+2.7%	0 → 0
BenchmarkEngineSetContainsMiss	167 → 167	-0.2%	0 → 0
BenchmarkEngineSetMembers	6638 → 6547	-1.4%	66 → 66
BenchmarkEngineSetRemove	1966 → 1809	-8.0%	1 → 1
BenchmarkEngineTopKAdd	1586 → 1550	-2.3%	21 → 21
BenchmarkEngineTopKAddParallel	1739 → 1740	+0.1%	21 → 21
BenchmarkEngineTopKListHit	621 → 616	-0.9%	11 → 11
BenchmarkEngineTopKListHitParallel	703 → 707	+0.5%	11 → 11
BenchmarkEngineZAdd	12629 → 12116	-4.1%	8 → 8
BenchmarkEngineZCard	153 → 151	-1.3%	0 → 0
BenchmarkEngineZRange	183072 → 183732	+0.4%	1017 → 1017
BenchmarkEngineZRangeByScore	183402 → 183141	-0.1%	1021 → 1021
BenchmarkEngineZScoreHit	100 → 99	-1.7%	0 → 0
BenchmarkStoreBitSet	107 → 106	-0.1%	1 → 1
BenchmarkStoreCGetHit	68 → 68	-0.0%	0 → 0
BenchmarkStoreCIncr	94 → 95	+0.1%	1 → 1
BenchmarkStoreCMSIncr	109 → 111	+1.5%	0 → 0
BenchmarkStoreCMSIncrN	109 → 109	+0.1%	0 → 0
BenchmarkStoreGeoAdd	12255 → 12007	-2.0%	2 → 2
BenchmarkStoreGeoPosHit	78 → 86	+10.4%	0 → 0
BenchmarkStoreGeoPosMiss	79 → 81	+1.9%	0 → 0
BenchmarkStoreGetHit	297 → 298	+0.5%	2 → 2
BenchmarkStoreGetHitParallel	368 → 355	-3.7%	2 → 2
BenchmarkStoreHGetHit	85 → 87	+2.4%	1 → 1
BenchmarkStoreHGetMiss	80 → 80	+0.8%	0 → 0
BenchmarkStoreHLLAdd	95 → 95	+0.5%	0 → 0
BenchmarkStoreHSet	12962 → 13024	+0.5%	3 → 3
BenchmarkStoreJsonSet	26200 → 26662	+1.8%	207 → 207
BenchmarkStorePut	307 → 305	-0.7%	2 → 2
BenchmarkStorePutParallel	392 → 338	-13.7%	2 → 2
BenchmarkStoreSetAdd	11644 → 11374	-2.3%	2 → 2
BenchmarkStoreSetContainsHit	75 → 79	+4.4%	0 → 0
BenchmarkStoreSetContainsMiss	76 → 79	+4.2%	0 → 0
BenchmarkStoreTopKAdd	1030 → 1039	+0.9%	20 → 20
BenchmarkStoreZAdd	11950 → 11908	-0.4%	2 → 2
BenchmarkStoreZScoreHit	77 → 83	+7.7%	0 → 0
BenchmarkStoreZScoreMiss	77 → 81	+4.5%	0 → 0

_{Posted by CI bench job. Not a merge gate.}

[Code0987]

Opened too early — you wanted to review first. Branch feat/refactor-hll kept.

[Code0987]

Closing — you asked to review locally, not commit/PR.