fix(deps): audit all CI extras and upgrade vulnerable packages

[mldangelo-oai]

Summary

• Audit the same all-ci dependency graph CI installs, including every pinned transitive dependency without launching another resolver.
• Update aiohttp to 3.14.3, cryptography to 50.0.0, GitPython to 3.1.58, and Keras to 3.15.0 to address 19 open Dependabot alerts.
• Replace the full MLflow server with its already-present secure mlflow-skinny tracking client while retaining every registry and artifact API ModelAudit imports.
• Add regression floors for all four packages, require the secure tracking client in every MLflow extra, and enforce complete CI audit coverage.

Why these changes belong together

The individual aiohttp and cryptography dependency PRs deadlock because each leaves the other's audit failures. Upgrading cryptography while retaining full MLflow forces MLflow 3.2.0 and PyArrow 21.0.0, introducing 27 vulnerabilities the current base-only audit misses. mlflow-skinny already exists in the current dependency graph and contains every API the integration uses, allowing patched cryptography without that vulnerable server downgrade or a new package dependency.

Validation

• Before: new version-floor regressions failed for all four current lock entries; the insecure MLflow fallback failed its tracking-client and audit-coverage regressions.
• After: dependency, workflow, and release suite: 157 passed; one unrelated macOS-host invalid-UTF8 path test was deselected.
• Actual skinny-only MLflow integration: 178 passed, 1 skipped; skinny-only MLflow CLI: 12 passed.
• Nightly-adjacent constructor and Joblib regressions: 34 passed.
• Broader macOS fast-suite smoke run: 3,411 passed before three known cache failures on unchanged code; those Darwin ancestor-churn failures are separately addressed by fix(cache): preserve macOS entries during ancestor churn #1800.
• Current all-ci and all-extras pinned dependency graphs both report No known vulnerabilities found.
• Full-repository Ruff lint and format, changed-file mypy, and changed Markdown/YAML Prettier checks passed.
• Full local mypy reproduces 14 pre-existing errors in six unchanged baseline files; GitHub Actions will verify Linux lock consistency and the complete CI matrix on this exact head.

[mldangelo-oai]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. 🚀

Reviewed commit: 5dce549176

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".

[github-actions]

Workflow run and artifacts

Performance Benchmarks

Compared 13 shared benchmarks with a regression threshold of 15%.
Status: 0 regressions, 0 improved, 13 stable, 0 new, 0 missing.
Aggregate shared-benchmark median: 4.369s -> 4.374s (+0.1%).

Workload	Benchmark	Target	Size	Files	Baseline	Current	Change	Status
nested-payload-review	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_nested_payload_review[nested_raw]	nested_raw	78 B	1	269.4us	284.1us	+5.5%	stable
padded-multi-stream-upload	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_padded_multi_stream_upload	multi_stream_padded	4.1 KiB	1	330.8us	344.1us	+4.0%	stable
nested-payload-review	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_nested_payload_review[nested_base64]	nested_base64	98 B	1	300.3us	295.6us	-1.5%	stable
mixed-model-repository	tests/benchmarks/test_scan_benchmarks.py::test_scan_release_candidate_repository	release-candidate	547.3 KiB	32	633.04ms	640.45ms	+1.2%	stable
duplicate-heavy-registry	tests/benchmarks/test_scan_benchmarks.py::test_scan_duplicate_registry_snapshot	registry-snapshot	915.2 KiB	13	578.34ms	574.57ms	-0.7%	stable
clean-training-checkpoint	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_clean_training_checkpoint	safe_large	278.2 KiB	1	110.76ms	110.13ms	-0.6%	stable
direct-malicious-upload	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_direct_malicious_upload	malicious_reduce	52 B	1	228.4us	227.4us	-0.5%	stable
suspicious-pickle-intake	tests/benchmarks/test_scan_benchmarks.py::test_scan_suspicious_pickle_intake	suspicious-intake	183.8 KiB	4	149.93ms	150.47ms	+0.4%	stable
warm-cache-rescan	tests/benchmarks/test_scan_benchmarks.py::test_scan_warm_cached_repository_rescan	release-candidate	547.3 KiB	32	155.50ms	155.03ms	-0.3%	stable
single-checkpoint-preflight	tests/benchmarks/test_scan_benchmarks.py::test_scan_single_checkpoint_before_load	single_checkpoint.pkl	183.0 KiB	1	104.26ms	104.52ms	+0.3%	stable
chunked-upload-stream	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_chunked_upload_stream	chunked_stream	278.2 KiB	1	113.50ms	113.23ms	-0.2%	stable
rejected-basic-auth-candidates	tests/benchmarks/test_scan_benchmarks.py::test_rejected_basic_auth_candidates_scan_linearly	-	371.1 KiB	1	2.522s	2.525s	+0.1%	stable
nested-payload-review	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_nested_payload_review[nested_hex]	nested_hex	130 B	1	305.4us	305.6us	+0.1%	stable

[chatgpt-codex-connector]

Security review completed. No security issues were found in this pull request.

Reviewed commit: 5dce549176

View security finding report

Only the user who started this review can view the report in Codex.

ℹ️ About Codex security reviews in GitHub

This is an experimental Codex feature. Security reviews are triggered when:

• You comment "@codex security review"
• A regular code review gets triggered (for example, "@codex review" or when a PR is opened), and you’re opted in so security review runs alongside code review

Once complete, Codex will leave suggestions, or a comment if no findings are found.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 76.29%. Comparing base (c29586b) to head (5dce549).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1808   +/-   ##
=======================================
  Coverage   76.29%   76.29%           
=======================================
  Files         217      217           
  Lines      115676   115676           
  Branches    27245    27245           
=======================================
+ Hits        88256    88257    +1     
  Misses      20023    20023           
+ Partials     7397     7396    -1     

see 1 file with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.