Extract status and doctor coverage from the monolithic CLI integration target

[errfld]

Summary

Extract the status and doctor end-to-end coverage from the monolithic CLI integration test target into a focused diagnostics integration test module, while preserving all existing behavior and platform coverage.

Concrete file/function evidence

• crates/git-smee-cli/tests/cli_integration.rs is 2,226 lines and contains coverage for every CLI command plus Unix/Windows hook execution and config-path edge cases.
• Diagnostics alone form a cohesive block around cli_integration.rs:210-532: healthy doctor text/JSON, status coverage/drift text/JSON, malformed/missing config, unmanaged hooks, marker placement, custom hooks paths, and stale managed hooks.
• The production code now has focused crates/git-smee-cli/src/status.rs, doctor.rs, and diagnostics.rs modules, but their end-to-end tests remain interleaved with install, run, init, migrate-hooks, and platform wrapper scenarios.
• crates/git-smee-cli/tests/common.rs is only 68 lines and already provides the shared TestRepo fixture, leaving a clear seam for multiple integration test targets.

Architectural concern

Production diagnostics have cohesive ownership after issue #164, but the integration-test extension path does not mirror it. Adding a status or doctor contract case requires navigating a 2,000+ line all-commands target, and unrelated command scenarios share one compilation/test unit. This makes diagnostics contract changes harder for a human reviewer to locate and increases collision risk as command coverage grows.

Proposed direction

Move only the status/doctor endpoint-style CLI tests into a focused target such as crates/git-smee-cli/tests/diagnostics_integration.rs. Keep reusable repository/config/hook setup in tests/common.rs; move diagnostics-only helpers with the new target. Preserve test bodies and assertions mechanically before considering any cleanup.

This is a test-ownership refactor, not a production behavior change or a request to reorganize every integration test in one PR.

Acceptance criteria

• Existing status and doctor text/JSON, drift, config-error, hooks-path, unmanaged-hook, and stale-wrapper scenarios run from a focused diagnostics integration test target.
• cli_integration.rs no longer contains those diagnostics scenarios.
• Shared fixtures live in tests/common.rs without command-specific assertions leaking into the common module.
• No test is deleted, weakened, ignored, or loses its existing cfg platform behavior.
• Production source and public CLI output remain unchanged.

Validation expectations

• cargo fmt --all -- --check
• cargo clippy -p git-smee-cli --all-targets --all-features -- -D warnings
• Run the new diagnostics target directly (for example cargo test -p git-smee-cli --test diagnostics_integration).
• Run cargo test -p git-smee-cli --all-targets --all-features to prove the split did not drop or duplicate coverage.

[errfld]

Implementation update for #197

Summary of changes

• Extracted all 14 status/doctor CLI scenarios into crates/git-smee-cli/tests/diagnostics_integration.rs.
• Removed those scenarios from the monolithic cli_integration target without changing their bodies, assertions, or platform conditions.
• Kept reusable setup in tests/common.rs; added a scoped dead-code allowance because each Cargo integration target compiles the shared fixture independently and uses a different subset.

Preflight knowledge checks

• Memory search run: no matching memory issues for CLI diagnostics/status/doctor integration coverage.
• Decision search run: no matching decision issues.
• Blocker/parent/linked issues reviewed: no blockers, dependencies, or sub-issues; issue Split CLI diagnostics and command orchestration into focused modules #164 was reviewed as the completed production-module predecessor.
• Prior task notes reviewed: issue Extract status and doctor coverage from the monolithic CLI integration target #197 had no prior comments; recent architectural-review implementation history was reviewed.

Validation

• TDD RED: cargo test -p git-smee-cli --test diagnostics_integration failed before extraction because the target did not exist.
• Baseline: cargo test -p git-smee-cli --test cli_integration passed 86 tests before extraction.
• Focused GREEN: diagnostics target passed 14 tests; remaining CLI target passed 72 tests.
• Package GREEN: cargo test -p git-smee-cli --all-targets --all-features.
• Lint/format GREEN: cargo clippy -p git-smee-cli --all-targets --all-features -- -D warnings; cargo fmt --all -- --check.
• Structural parity: all 14 moved test bodies exactly match origin/main, are absent from the old target, and occur once in the new target.
• Hosted validation: PR test(cli): split diagnostics integration coverage #198 CI matrix, MSRV, Linux/macOS/Windows, lint, coverage, security, repository guards, workflow lint, CodeQL, and CodeRabbit all passed on exact head fc7a5509959c23256311ec2c4b5e24c188023ac4; CodeRabbit approved with no inline findings.

Learnings

• Shared integration fixtures need target-local lint tolerance when focused Cargo test targets intentionally consume different fixture subsets.
• A mechanical source-parity assertion is a useful guard for test-ownership refactors: the split preserved the original 86 scenarios as 72 + 14.

Follow-ups

• No follow-up issue needed; this focused split fully satisfies Extract status and doctor coverage from the monolithic CLI integration target #197.

References

• PR: test(cli): split diagnostics integration coverage #198
• Commit: fc7a5509959c23256311ec2c4b5e24c188023ac4

[errfld]

Merged PR #198 as 1cbe6610491ab7fcef42d5d2751c91156de3b27a after all hosted CI/CodeQL checks passed and CodeRabbit approved the exact PR head. This completes the diagnostics integration-test ownership split; no follow-up slices remain.