Add imara-diff v0.2 with slider heuristics support

[Copilot]

Adds imara-diff v0.2 alongside existing v0.1 to provide Git's slider heuristics for more intuitive diff placement.

Changes

• Dependency: Added imara-diff-v0_2 (v0.2.0) alongside existing v0.1.8
• API: Exposed v0.2 types via blob::v0_2 module (Diff, UnifiedDiffPrinter, IndentHeuristic, etc.)
• Convenience function: Added diff_with_slider_heuristics() that computes and postprocesses diffs in one call
• Tests: Added 10 tests demonstrating UnifiedDiffPrinter usage, custom heuristics, and hunk iteration

Usage

use gix_diff::blob::{diff_with_slider_heuristics, v0_2::{Algorithm, InternedInput}};

let input = InternedInput::new(before, after);
let diff = diff_with_slider_heuristics(Algorithm::Histogram, &input);

// Generate unified diff output
let printer = v0_2::BasicLineDiffPrinter(&input.interner);
let output = diff.unified_diff(&printer, v0_2::UnifiedDiffConfig::default(), &input);

Maintains full backward compatibility with existing v0.1 API.

Original prompt

In gix_diff::blob uses imara-diff v0.1 to provide gix_diff::blob::diff() for callers that don't need Git slider heuristics. Add the v0.2 of imara-diff as dependency and expose its Diff structure via git::blob::diff_with_slider_heuristics(). Add tests for it as well as you see fit, showing how the UnifiedDiffPrinter can be used.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Christoph Rüßler (cruessler)]

Sebastian Thiel (@Byron) Counting removals and additions probably is not the best example to showcase slider heuristics as line numbers are not affected by moving sliders up and down. :-)

[Sebastian Thiel (Byron)]

Alright, Christoph Rüßler (@cruessler), I have worked it over and think v2 is good as experimental feature to play around with. The integration tests are a test-bed on its own, but I think now it's more straightforward to try it with gix-blame for example.

Performance would be interesting, and the usability of the unified diff printer, which desperately misses &[u8] compatibility (and a lot of tests). I have also created pascalkuthe/imara-diff#41 to attack this from another angle, as V2 is quite underdocumented still.

I hope this helps 🙏