[tae898]

Refs #555. Docs sweep for statements that today's changes (26.11.1 released, the pass order, the io fix) made false or incomplete. Docs only, nothing under benchmarks/experiments/*.py or the runner path changes, so it is safe to merge during the pass (docs-only merges are allowed).

File	Before	After
README.md, bindings/python/README.md, bindings/python/docs/index.md (banner)	"The first releases of the new packages are 26.11.1, starting as pre-releases (uv add --pre arcadedb). ... arcadedb-embedded will get a last release that installs arcadedb."	"Both packages are released as 26.11.1 (uv add arcadedb). ... the last release of arcadedb-embedded, 26.11.1, installs arcadedb." PyPI shows arcadedb, arcadedb-jvm and arcadedb-embedded all at 26.11.1
same three files	"This repository will be archived; the benchmark tooling stays here for now."	"This repository will be archived once the benchmark tooling moves to humemai/dbbench (planned for mid-November 2026, after the last measuring pass); until then the tooling stays here." (same facts as AGENTS.md)
README.md	install block uv add --pre arcadedb / uv add --pre arcadedb-jvm, comment "until 26.11.1 is a final release"; PyPI badge with include_prereleases	uv add arcadedb / uv add arcadedb-jvm, no --pre, badge without include_prereleases
bindings/python/README.md	"new projects use: uv add --pre arcadedb"	"new projects use: uv add arcadedb"
benchmarks/README.md	"report five repetitions per cell"	"report up to five repetitions per cell (n=3 for every engine first, then n=4, then n=5)"
benchmarks/experiments/PAGE-SPEC.md (rule 2)	"Five repetitions per cell, reached in two passes"	"reached in whole passes: n=3 for every engine first (the four ArcadeDB arms included), then n=4, then n=5"
benchmarks/experiments/PROTOCOL.md (N=5 rule)	only the 26.10.1 two-pass story	adds the 2026-11 order: n=3 for every engine including the four ArcadeDB arms, then n=4, then n=5, each a whole pass; comparators n=3 for sure, n=5 where passes 4 and 5 finish before the freeze
benchmarks/experiments/CAMPAIGN.md (paper profile)	order: pass 3, re-runs, four-arm pass, passes 4 and 5	adds the 2026-11-05 amendment (DECISIONS #232): when the finals are on PyPI and the four-arm runner is deployed at REORDER reruns finished, the four-arm pass starts first and the window re-runs follow it, before PASS45.go; otherwise the old order stands
PROTOCOL.md (IO per phase), CAMPAIGN.md (IO)	no mention of the doubled figures	mini rows before the #555 fix count the stacked NVMe 259:0 and device-mapper 252:0 twice, so io_*_mib is about 2x; the fix (#570) counts a stacked device once and stamps io_stat_dedup: 1; the freeze halves mini rows without the stamp in runs_paper.csv only (#571), runs.jsonl keeps raw values, generated/io_stat_correction.json lists them

Checked and left alone: PUBLISHING.md already names generated/io_stat_correction.json and the io_stat_dedup rule (#571) and is correct; benchmarks/python-bindings/README.md and its tables say five repetitions, which is true for the frozen SciPy artifact; FAIRNESS.md F4 ("If ArcadeDB gets five passes ... so does every comparator") is a rule, still true; provenance_check doc line "3 to 5 repetitions" is true. cypherglot and humemdb appear only in three code comments (export_web.py, graph_common.py, runner.py), no doc: nothing to change, and both repos are already archived on GitHub. The PROTOCOL io paragraph describes #570 (still a draft, merge held) as the fix; it says what the fix does, not that it has landed.

pre-commit on the changed files: clean, no rewrites.

🤖 Generated with Claude Code