deps(gazelle): upgrade rules_go to remove patching of tree-sitter

[jbedard]

Update rules_go to include bazel-contrib/rules_go#4298
Update gazelle to align with the version the rules_go bzlmod will bring in, and ensure the go.mod version is the same as bzlmod version.
Update go to 1.21 to include the slices library that some of the go.mod updates depend on.

Fixes#2956.

[dougthor42]

Odd... how did you avoid the windows issues with go-tree-sitter that I was seeing in #2962?

Ah well, I'm not going to look a gift horse in the mouth. This is awesome, thanks!

Requests:

1. Update CHANGELOG.md to specify the version upgrades and that load statements might get reordered.
2. Update MODULE.bazel's rules_go version to match what's used by gazelle/MODULE.bazel.
3. Update internal_dev_deps.bzl to match too.
4. Update examples/(bzlmod_)build_file_generation to match versions too.

[jbedard]

Odd... how did you avoid the windows issues with go-tree-sitter that I was seeing in #2962?

I have no idea, don't jinx it! 😅

[jbedard]

Now after upgrading things there's some odd failures. @dougthor42 can you take a look?

[dougthor42]

To summarize all the comments:

• You should be able to revert the .bazelignore and .bazelrc changes.
  • Doing so tested just fine for me locally, so 🤞.
• You may be able to revert the change to root MODULE.bazel and WORKSPACE
  • Sorry for requesting those initially!
• It seems like ae0f4be worked and just needs a CI retry.
• Some nits

[dougthor42]

LGTM; I'm just going to test it on our internal monorepo this morning to double check, but I don't expect any issues.

[dougthor42]

I opened #3180 for the CI flake.

I'm just going to test it on our internal monorepo this morning to double check, but I don't expect any issues.

Things look good on my end. Approved.

[aignas]

Shall we try merging this or shall we hold off because we want to investigate the Windows flake?

[dougthor42]

@rickeylev has asked Florian internally to give me access to buildkite and the Bazel org so I can retry builds, so I was waiting for that.

My plan is to try to assess just how flaky things are - 50%, 10%, etc. - and then then merge if (a) the flakes are unrelated to this PR or (b) the flakes are "relatively rare"¹

Footnotes

1. for some unknown definition of "relatively" and "rare". ↩

[rickeylev]

I'd be fine with merging now.