[Bot] Update dependencies - #3009

@mayeut, was there a problem with upload?

🤖 AI text below 🤖

#3009 is incomplete. Most Docker pins still point to the old 2026.10.03-1 images, which have Python 3.15.0rc3. The CI tests are still running, so they don't tell us anything yet.

What is correct in the PR:

• macOS, Windows, Android and iOS: the cp315 downloads now use 3.15.0 final. The iOS support packages for 3.13 and 3.14 also moved to b16 and b12.
• python-build-standalone: updated to 20261009, which includes 3.15.0.
• Other updates: virtualenv went to 21.14.6, and the constraints files have small version bumps.
• docs/working-examples.md: the lines changed but look the same, so this is probably whitespace only. It does no harm.

What did not update. These pins are still on 2026.10.03-1:

• manylinux_2_28, for all architectures, including pypy.
• musllinux_1_2, for all architectures.
• manylinux_2_39 riscv64.

Only manylinux2014, manylinux_2_31, manylinux_2_34 and manylinux_2_35 moved to 2026.10.09-1.

Cause: the problem is in the manylinux release, not in our script. bin/update_docker.py reads the latest tag of each multi-arch repo (for example quay.io/pypa/manylinux_2_28). The "Deploy multi-arch images" job in manylinux run 37987958535 had errors when it pushed to quay:

• manylinux_2_28:latest: the bearer token request failed. The 2026.10.09-1 tag exists, but latest was not moved to it.
• manylinux_2_39:2026.10.09-1: the blob check failed, so the job skipped latest.
• musllinux_1_2:2026.10.09-1: quay returned 502 Bad Gateway, so the job skipped latest.

The job still reported success, so nothing flagged the failed pushes. The per-architecture repos, such as manylinux_2_28_x86_64, all have 2026.10.09-1.

Next step: someone has to re-run that deploy job in manylinux, or tell mayeut on #1883. After that, run the update workflow again. The other way is to pin tag="2026.10.09-1" in bin/update_docker.py for those three images for now. I recommend the re-run.

was there a problem with upload?

Indeed there was. It's been re-run and the workflow will be fixed.
Every image has been updated, I re-triggered the update workflow here.