(feat)ci: pypy3.11 support

[kclowes]

🗒️ Description

Allows this repo to run using pypy3.11. The new CI job runs in ~15m, which is generally ~30s-1m slower than our next slowest job. I'm happy to spend more time trying to get it down, but didn't know how much priority to give it here.

🔗 Related Issues or PRs

#1839

✅ Checklist

• All: Ran fast tox checks to avoid unnecessary CI fails, see also Code Standards and Enabling Pre-commit Checks:

  uvx --with=tox-uv tox -e lint,typecheck,spellcheck,markdownlint

• All: PR title adheres to the repo standard - it will be used as the squash commit message and should start type(scope):.
• All: Considered adding an entry to CHANGELOG.md.
• All: Considered updating the online docs in the ./docs/ directory.
• All: Set appropriate labels for the changes (only maintainers can apply labels).

[danceratopz]

Amazing! Really nice to have this!

As a sanity check could you please:

1. Compare cpython and pypy 3.11 fill run times on this branch and cpython on main?
2. Diff the generated fixtures with pypy on this branch to fixtures generated on main? We use the hasher utitlity to diff? This should be adequate:

 uv run fill -m "not slow and not benchmark" -n auto

I think the docs are a bit borked on this, but this comment explains how to do this:
#792 (review)

I absolutely recommend doing this on the hive server for speed!

[kclowes]

As a sanity check could you please:

1. Compare cpython and pypy 3.11 fill run times on this branch and cpython on main?

Before Osaka tests became the default, this was working really well. Python 3.12 on main was able to fill tests in 8:49 on the hive-server, and pypy3.11 on this branch was running in 4:48. However, after the Osaka tests became the default, locally and on hive-server I am erroring out. I'm not seeing the same thing on CI, so I am not sure what's going on. It may be worth digging in, but it seems like this isn't a huge priority right now. I took the pypy fill out of CI because it's taking more than an hour to run right now, and EELS is already testing the fill command on their end. Happy to keep looking here or trying to get the test times down if that becomes a priority.

2. Diff the generated fixtures with pypy on this branch to fixtures generated on main? We use the hasher utitlity to diff? This should be adequate:

The hashes do match!

[marioevz]

LGTM! Just one question but it's not a blocker. Thanks!