fix(test): make clone stage dominance independent of real wall-time races

[DecisionNerd]

Fixes #1650.

Problem

hub_clone::tests::both_identity_forms_import_and_reopen_the_same_real_project injected a two-second thread::sleep into one stage per case and asserted that stage had the largest real wall time. Real import, filesystem work, and scheduling have no two-second bound, so under load PortableImport could out-dominate the injected NetworkTransport stage. That's an invalid timing assumption, not a clone defect. The failure was observed while validating #1623.

Change

• Add a CloneClock (crates/graphforge-cli/src/hub_clone.rs). Production code always uses Monotonic, which keeps the unchanged Instant::now()/thread::sleep behavior. The Manual variant is #[cfg(test)] only.
• The five injected-work attribution cases run on an isolated manual clock that only the injected work advances. They still perform real clones, imports, receipts, handoffs and facade reopens.
• Assertions are tightened, not weakened:
  • the expected dominant component matrix is unchanged;
  • the dominant stage must equal exactly 2_000_000_000 ns;
  • stages must not overlap or reverse;
  • finished_ns must equal the end of the last stage.
• No sleeps were increased, no retries or serialization were added, and no assertions were removed.

Validation

• cargo fmt --all -- --check: clean.
• cargo clippy -p graphforge-cli -- -D warnings, matching the CI Rust Quality gate command: clean. With --all-targets there are pre-existing lint errors in unrelated test files (maintenance_cli/tests.rs, research_cli/interchange.rs, tests/{branches,research_journey,slices}.rs) that also fail on main. None are in hub_clone.rs.
• python3 scripts/test_environment.py -- cargo test -p graphforge-cli --lib hub_clone: 21 passed, 0 failed, 1 ignored, in 0.36 s. The attribution cases no longer spend about 10 s of real sleeps.

🤖 Generated with Claude Code

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository: CurateLabs/graphforge/.coderabbit.yaml

Review profile: CHILL

Plan: Advanced

Run ID: b099d45b-7ba3-4221-ab70-721cadd7a73d

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

Warning

Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[DecisionNerd]

Independent review and local validation for the #1623 blocker:

• make pre-push passed at 6231fff43faeb3a1987841583d971d611ac1a880 after refreshing onto the then-current main. Native Rust coverage, rebuilt Python/Node bindings, wrapper coverage, and final thresholds passed (/tmp/gf-1650-prepush-refreshed.log). Rust quality reused compatible validated evidence.
• The current PR code diff at a6631ab12abe126c4f232d76a8dd6164d5005967 is identical to that independently reviewed change; only its main integration advanced. It still changes only hub_clone.rs.
• The expected dominant-component matrix, real clone/import/reopen proof, and fail-open equivalence remain. Manual attribution time is isolated; production monotonic timing is unchanged.
• Current exact-head CI Gate passed; mergeStateStatus is CLEAN, there are no review threads, and the sole closing reference is fix(test): make clone stage dominance independent of real wall-time races #1650.

The earlier concurrency lane failed during dependency-index transport before tests ran; its HTTP/2 framing error remains recorded in run 36659106567. The current refreshed-tree run completed the concurrency lane successfully.