Prepare the 2.17.0.dev4 release.

[kaos]

No description provided.

[kaos]

Internal (put these in a PR comment for review, not the release notes)

• Revert "Option required flag" (#18802)

• upgrade Rust to v1.69.0 (#18783)

• Support docformatter 1.6.0+ (#18790)

• Add new required flag for options. (#18780)

• Upgrade mheap/github-action-required-labels to v4.0.0 (#18778)

• Use a released version of bollard (#18777)

• Get rid of spectral rust crate. (#18774)

• Remove superfluous custom helm tool resolves. (#18765)

• Switch to new lockfile generation script. (#18754)

• A new script to generate the built-in tool lockfiles. (#18751)

• Add a fingerprinted environment variables option to pyenv provider backend (#18708)

• Avoid using private type hashlib._Hash, which may not exist (#18740)

• Add missing backtick to docstring (#18735)

• Prefactor: Move get_hash from docker backend to strutil (#18715)

• Release script tweaks (#18728)

• Removes a deprecation warning whose removal was blocked by plugin code (#18722)

• Update Toolchain pants plugin to v0.27.0 (#18720)

• Add tool_id classproperty to LintRequest (#18711)

• Switch many tests to use PythonRuleRunner. (#18712)

• A RuleRunner subclass to set ICs in Python-related tests. (#18709)

[jsirois]

The real CI failure is here and terrifying: https://github.com/pantsbuild/pants/actions/runs/4779380793/jobs/8496439611?pr=18804#step:11:306

That does not fail on main and this PR does not disturb BUILD files or production or test code. The test is of parsing - there is no network involved or anything else that should be flaky. This leaves just 1 thing, reporting on an old cached result, and one that does not match. The logs do indicate a remote cache token is acquired :/

In fact the BuildSense link for that shard, and that shard alone OOPSes, not sure how seriously to take that coincidence: https://app.toolchain.com/organizations/pantsbuild/repos/pants/builds/pants_run_2023_04_23_16_46_28_583_fd49b5383daf4a0290e2a4de077b2459/type/goal/?subTab=test&testTab=RESULTS

All the terror aside, this PR itself looks good to me.

[kaos]

Ouch. Yea.. that's worth looking into. Q is how far we'll get if we can't reproduce the issue. Maybe if someone is clever enough to deduce what must've gone wrong in order to "force" a reproduction..?

[kaos]

FWIW I saw the same error again in CI on the re-run. Went away when running on a new commit (after merge with main).