docs: publish verified 4.8 installation and benchmark workflows

[jacksonpradolima]

The 4.8 integration is published and verified, but README/setup pages still advertise the 4.7 wheel or require building unreleased source. This updates the public entry points to the actual 4.8.0 wheel and tested workflows.

• Pin the published wheel SHA-256 712bfefe58d3eff09ef9adf239b32ac40594c3c80f17e5330c092f1b8f699b75 throughout README, setup, application API, observability and ETL guides. All 36 newly added installation URLs were checked against the actual asset; historical release snapshots retain their versions.
• Present both real CI workflows with concrete units/counts: LRTS method executions and RTPTorrent class executions remain distinct. Eight modeled rounds / default Random-UCB / 16 result rows are described alongside source admission and comparison limits.
• Make the ZIP64 and RTPTorrent Parquet guides usable without a source checkout. Download the exact release-tagged owned scripts with pinned SHA-256, then run twice offline. Clarify that scripts are separate repository files; the wheel contains recipes/owned fixtures.
• Mark Add bounded ZIP64 source selection for real benchmark archives #301/Deliver real RTPTorrent Parquet ETL and native replay #302 delivered via feat(corpus): add bounded ZIP64 source selection #314/feat(corpus): deliver real RTPTorrent Parquet ETL and native replay #315/feat(corpus): ship verified RTPTorrent Parquet and replay workflows #316, retain Build a diversified, versioned benchmark corpus for adaptive CI test prioritization #261 open and Expand LRTS source slices and resolve stage and temporal admission evidence #303–Freeze and execute the confirmatory corpus and publish the cross-corpus report #313 as remaining work. Regenerate llms text deterministically.

Documentation only. Code, tests, examples, configuration and lock are unchanged. No artifact migration or new package bump is requested; the existing versioned contracts, private Linux worker guard and source-admission limits retain their behavior. The release description remains the reviewed en-US notes automatically published from docs/releases/4.8.0.md.

Validation before this update:

• Release workflow succeeded. The automatic version commit changes only the self-version in pyproject.toml/uv.lock; reviewed source bytes remain identical.
• All six release assets match GitHub digests. Source/wheel/tag/lock contents agree; SBOM binds the exact wheel. Both Sigstore bundles verified offline against this repository's actual main-push release workflow identity/SHA.
• Fresh published base wheel outside checkout passed existing numeric/identity/import-boundary/owned workflows, ten actual corpus profiles twice, controlled Java source capture (95 cases per role; expected and negative outcomes), real LRTS conversion/native/checkpoint twice, and real RTPTorrent complete-archive conversion/native/checkpoint twice. The latter retains 46,995 observations / 1,662 jobs, 18 projected executions, 16 result rows and committed step 8.
• Exact tagged ZIP64 and Parquet example downloads/checksums passed, each executed twice with the published wheel outside checkout. MCP/telemetry/ClickHouse extra URLs resolve in a dry run; the base environment remains unchanged. Published-wheel Linux memory regressions: 10 passed.
• Strict isolated pip-audit of 238 locked registry packages: 0 known vulnerabilities. Open Dependabot/Code Scanning/advisories and pending bump PRs remain zero.

For this documentation diff, passed lint/format/Pyright/ty/docstrings/structural complexity/Pylint-E, strict docs, build, deterministic llms regeneration and git diff --check. Full code baseline already passed 2,542 tests, 1 platform skip, 95.31% coverage locally and complete Linux CI in #315/#316/main. The new head will pass its own CI/review before merge.

Local OpenCodeReview 1.12.13 delegation only:

npx --yes @alibaba-group/open-code-review@1.12.13 delegate preview \
  --format json --from main --to HEAD -b 'Refresh verified published4.8 docs and accurate evidence/admission limits'
npx --yes @alibaba-group/open-code-review@1.12.13 delegate preview --format json
npx --yes @alibaba-group/open-code-review@1.12.13 delegate rule \
  --format json --from main --to HEAD <all-15-changed-paths>

The delegate includes the navigation TOML and excludes all 14 Markdown/generated files. Read applicable rules and manually reviewed 15/15 files (100%), including installation/hash consistency, complete commands, links, data units, release status and generated llms diff. No confirmed critical/high/medium findings remain. Evidence: coleman-317-nav-ocr-main-head.json, coleman-317-nav-workspace.json, coleman-317-nav-ocr-rules.json, coleman-317-nav-complete-local-review.json; published proofs in coleman-4.8.0-final-delivery-proof.json and associated asset/signature/real-flow logs.

Follows #316; contributes to #261. No additional corpus admission, policy superiority, independent-fault or parallel-savings claim is introduced.

Review correction 80a3a2d exposes the published4.8 release first in site navigation. An actual rendered-primary-navigation check was RED on the original head (4.7 first) and GREEN after the correction (4.8 first,4.7 preserved). Repeated strict-doc build, fast quality/Pylint/build, generated-text check and complete15-file review passed. The last14 documentation file bytes and all core/test/example/version/lock bytes remain unchanged.

[jacksonpradolima]

@codex review

Please review documentation-only head 37bcbabd6d5784c3dc165ad6780af44474be0fec. It follows verified publication of4.8 (#316). Check actual artifact/example pins, complete standalone installation/CI flows, unit/count consistency, published-versus-planned status and preserved source-admission limits. All14 excluded Markdown/generated files were manually reviewed with local delegated rules; fast checks/Pylint/strict-docs/build passed. Signed release assets, automatic version/lock delta, actual published real LRTS/RTPTorrent/native/checkpoint workflows, tagged owned scripts and Linux guards passed; isolated audit238packages found0knownvulnerabilities. No code/tests/version/lock changes or new release is requested.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-16T23:30:26.807217Z	80a3a2d	Manual request

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 37bcbabd6d

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[jacksonpradolima]

@codex review

Please review exact final head 80a3a2d87a24b97001a73d0043822fa3ca45b244. The only prior finding (release navigation) is fixed with actual rendered-primary-navigation RED/GREEN validation and strict build. Repeated full15-file local review and fast gates passed; code/tests/examples/version/lock and the previous14 documentation files remain unchanged. Requesting a fresh clean review and complete CI before merge.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Breezy!

Reviewed commit: 80a3a2d87a

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".