test: fix timing races from most sleep-based async tests

[Alenar]

Content

This PR includes fixes to tests that used real-clock sleeps to order tasks or check delays, which made them flaky on loaded CI machines.

This problem was identified for one the mithril-client feedback test by the following failing hydra run: https://ci.iog.io/build/15142234/nixlog/1. I then searched other potential offender in the code base.

Details:

• client: the feedback tests now take turns through Notify handshakes instead of sleeps. They are renamed to *_task(s) since they use tasks, not threads.
• relay: the repeater delay tests use a paused clock and explicit time::advance calls to check the message is not repeated before the delay ends and is repeated once it does.
  The other repeater tests use a shorter delay so they run faster.
• aggregator: the signature processor stops once a registration is notified, with a timeout so the test can't hang, and the retry-delay test advances a paused clock to just before the delay ends.
  Both tests now fail if the processor stops on its own instead of passing silently.
• aggregator: rename the wait_delay_on_error_in_seconds parameter to wait_delay_on_error since it is a Duration.

Pre-submit checklist

• Branch
  • Tests are provided (if possible)
  • Commit sequence broadly makes sense
  • Key commits have useful messages
• PR
  • All check jobs of the CI have succeeded
  • Self-reviewed the diff
  • Useful pull request description
  • Reviewer requested
• Documentation
  • No new TODOs introduced

[github-actions]

Test Results

     5 files  ±0     226 suites  ±0   1h 0m 41s ⏱️ - 4m 4s
 3 998 tests ±0   3 998 ✅ ±0  0 💤 ±0  0 ❌ ±0 
12 635 runs  ±0  12 635 ✅ ±0  0 💤 ±0  0 ❌ ±0 

Results for commit 663b6a9. ± Comparison against base commit 4eb474c.

This pull request removes 3 and adds 3 tests. Note that renamed tests count towards both.

mithril-client ‑ feedback::tests::send_event_in_one_thread_and_receive_in_another_thread
mithril-client ‑ feedback::tests::send_event_multiple_thread
mithril-client ‑ feedback::tests::send_event_same_thread

mithril-client ‑ feedback::tests::send_event_in_one_task_and_receive_in_another_task
mithril-client ‑ feedback::tests::send_event_multiple_tasks
mithril-client ‑ feedback::tests::send_event_same_task

♻️ This comment has been updated with latest results.

[Copilot]

🟡 Changes recommended

A repeater test can pass its pre-delay assertion before the spawned task has run.

1 open finding

• Yield before advancing clock to verify repeat delay · New

What changed in this PR

This PR makes async tests less dependent on real-clock sleeps, reducing timing-related failures on loaded CI machines.

Changes:

• Replaces client feedback sleeps with task handshakes.
• Uses paused clocks to test relay and aggregator delays, and shortens other relay test delays.
• Renames an aggregator Duration parameter for clarity.

File	Description
mithril-relay/​src/​repeater.rs	Updates repeater timing tests.
mithril-relay/​Cargo.toml	Enables Tokio test utilities.
mithril-client/​src/​feedback.rs	Coordinates feedback tests with notifications.
mithril-aggregator/​src/​services/​signature_processor.rs	Updates processor timing tests and parameter naming.

🧠 Review effort: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[turmelclem]

LGTM 👍

[jpraynaud]

LGTM 👍