Switch to imara-diff 0.2 in gix-blame

[cruessler]

This PR changes gix-blame to always use imara-diff 0.2 through gix-diff/blame-experimental. It removes the blob-experimental feature flag in gix-blame as well as the blame-experimental feature flag in gix.

This PR does not yet change gix-diff itself yet. I’m in the process of preparing a separate PR for that switch as well, but I thought the transition was easier to do and review in smaller PRs. If you think this should be done in a single PR, though, I’ll be happy to incorporate the other changes into this PR as well. Let me know what you think!

[Copilot]

Pull request overview

This PR completes the transition of gix-blame to always use the imara-diff v0.2-backed implementation (via gix-diff’s experimental blob diff feature), and removes the now-obsolete feature flags that previously selected between v0.1 and v0.2.

Changes:

• Remove blame-experimental feature wiring from the workspace (Cargo.toml, gitoxide-core, gix) and stop running gix-blame tests with the old blob-experimental flag.
• Make gix-blame always enable gix-diff’s blob-experimental feature and remove the gix-blame-level blob-experimental feature.
• Simplify/adjust the blame disparity test to run both Myers and Histogram under the unified setup.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
justfile	Stops running gix-blame unit tests with the removed blob-experimental feature.
gix/Cargo.toml	Removes the blame-experimental feature flag from gix.
gix-blame/tests/blame.rs	Updates the disparity regression test to run both algorithms under the always-v0.2 setup.
gix-blame/src/file/function.rs	Removes the non-experimental blob diff code path, leaving the v0.2-based implementation.
gix-blame/Cargo.toml	Removes the crate feature and enables gix-diff’s blob-experimental unconditionally.
gitoxide-core/Cargo.toml	Removes the blame-experimental feature passthrough.
Cargo.toml	Removes the workspace-level blame-experimental feature passthrough.

Comments suppressed due to low confidence (2)

gix-blame/tests/blame.rs:325

• This test uses multiple .unwrap() calls (e.g., Fixture::new().unwrap()), which makes failures harder to diagnose and goes against the project’s general “no unwrap” practice. Consider returning gix_testtools::Result from the test and using ?, or use .expect("…") with context.

        let Fixture {
            odb,
            mut resource_cache,
            suspect,
        } = Fixture::new().unwrap();

gix-blame/tests/blame.rs:349

• More .unwrap()s here (gix_blame::file(...).unwrap(), fixture_path().unwrap(), Baseline::collect(...).unwrap()). Prefer propagating errors via ? (by making the test return gix_testtools::Result) or at least adding contextual .expect() messages so failures are actionable.

        let lines_blamed = gix_blame::file(
            &odb,
            suspect,
            None,
            &mut resource_cache,
            source_file_name.as_ref(),
            gix_blame::Options {
                diff_algorithm,
                ranges: BlameRanges::default(),
                since: None,
                rewrites: Some(gix_diff::Rewrites::default()),
                debug_track_path: false,
            },
        )
        .unwrap()
        .entries;

        assert_eq!(lines_blamed.len(), 5);

        let git_dir = fixture_path().unwrap().join(".git");
        let baseline = Baseline::collect(git_dir.join(format!("{case}.baseline")), source_file_name).unwrap();

[Byron]

Fantastic, thanks!

Now we have experimental support by default for everyone who uses default gix features, but that's alright.