feat: cargo build/test/clippy with compact output

[FlorianBruniaux]

Summary

• rtk cargo build: strip Compiling/Downloading lines, show errors + summary
• rtk cargo test: show failures only + summary line
• rtk cargo clippy: group warnings by lint rule with locations

Files changed

• src/cargo_cmd.rs — new module with CargoCommand enum + 3 filter functions
• src/main.rs — Commands::Cargo, CargoCommands enum, match arms

Test plan

• 6 unit tests in cargo_cmd.rs (build success/errors, test pass/fail, clippy clean/warnings)
• cargo build compiles clean
• cargo clippy — no new warnings

Dependencies

Depends on #28 (shared infrastructure). Stacked PR series: PR 2/6

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Adds a new rtk cargo command group that runs cargo build/test/clippy and post-processes their output into a more compact, LLM-friendly format.

Changes:

• Introduce src/cargo_cmd.rs with CargoCommand plus output filters for build/test/clippy (with unit tests).
• Wire rtk cargo build|test|clippy into the CLI in src/main.rs.
• Expose/extend shared helpers used across commands (JSON string filtering, compact diff visibility, confirmation + package manager helpers).

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/cargo_cmd.rs	New cargo subcommand runner + output compaction filters and unit tests.
src/main.rs	Adds Commands::Cargo and dispatch to cargo_cmd::run.
src/utils.rs	Adds ok_confirmation, detect_package_manager, package_manager_exec and tests.
src/json_cmd.rs	Extracts filter_json_string() and updates run() to use it.
src/git.rs	Makes compact_diff pub(crate) for cross-module reuse.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

In filter_cargo_test, the test result: line is pushed into summary_lines inside the in_failure_section branch, and then it can be pushed again by the later if !in_failure_section && line.starts_with("test result:") block in the same loop iteration. This duplicates the summary line for failing test runs. Consider continue-ing after handling test result: in the failure-section branch, or refactor so the summary capture happens in exactly one place.

Suggested change

	summary_lines.push(line.to_string());
	summary_lines.push(line.to_string());
	continue;

[Copilot]

json_cmd::run used to include the filename in the JSON parse error context (e.g. "Failed to parse JSON: "). After switching to filter_json_string, parse failures lose that file context. Consider wrapping the filter_json_string call with with_context(|| format!("Failed to parse JSON: {}", file.display())) so file-based usage remains debuggable while keeping the generic error inside filter_json_string for piped input.

Suggested change

	let schema = filter_json_string(&content, max_depth)?;
	let schema = filter_json_string(&content, max_depth)
	.with_context(|| format!("Failed to parse JSON: {}", file.display()))?;

[Copilot]

test_detect_package_manager_default doesn’t actually verify the intended default behavior: it asserts the return value is one of ["pnpm", "yarn", "npm"], which will always pass regardless of lockfiles. To make this test meaningful/stable, run it in a temp directory without lockfiles (e.g., using tempfile + set_current_dir) and assert it returns "npm".

[Copilot]

PR description lists only src/cargo_cmd.rs and src/main.rs as changed, but this PR also modifies src/utils.rs, src/json_cmd.rs, and src/git.rs in the diff shown here. If this is due to the stacked PR dependency on #28, it would help to explicitly call that out (or rebase onto the merged base PR) so reviewers can distinguish the incremental changes from the shared-infra changes.