Make test_averaging_cancel more robust

[mryab]

Currently, test_averaging_cancel is one of the few flaky tests in our codebase: see e.g. https://github.com/learning-at-home/hivemind/actions/runs/14024261668/job/40777059821, where it failed 3 times out of 5.

Based on my investigation and discussion with @justheuristic, it looks like the problem is that it's very loosely defined: we start 4 averaging peers, cancel 2, and expect that the group size is 2 nonetheless. However, due to target_group_size=None by default, it's possible for the groups to have sizes of 3 and 1, which means that the len(control.result()) == 2 will occasionally fail.

To fix this issue, the PR introduces a separate test case with a fixed target_group_size. To make sure that peers don't average before cancellation, it also uses the averaging triggers to arrange the groups without starting the averaging itself

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 85.45%. Comparing base (d20e810) to head (ba2e073).
Report is 24 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #650      +/-   ##
==========================================
+ Coverage   85.39%   85.45%   +0.05%     
==========================================
  Files          81       96      +15     
  Lines        8006     8575     +569     
==========================================
+ Hits         6837     7328     +491     
- Misses       1169     1247      +78     

see 35 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[justheuristic]

Thanks! / LGTM