fix: accept destination dev-branch ancestors in integration history check (PTFE-3343)

[delthas]

Problem

Bert-E raised a false History mismatch (BranchHistoryMismatch, code 113) — telling the user to reset even though nothing was rebased — when an integration branch sat on a development branch that had not diverged from its parent (development/X.Y freshly branched off development/X.(Y-1), still pointing at the same commit D) and that integration branch carried a real merge commit.

When it happens

The trigger is a real (non-fast-forward) merge commit on w/X.Y whose first parent is the shared dev tip D. Two ways to get there:

• Automated cascade, "late" PR — when the PR's feature branch was branched before D (the PR is behind the dev tip), the auto-created w/X.Y merge cannot fast-forward (feature and D are siblings), so Bert-E produces a real merge commit whose first parent is D. PRs are frequently behind the dev tip, so this is the common case. (Thanks @tcarmet for spotting this path.)
• Manual resolution — an operator hand-resolves w/X.Y with a forced merge commit (e.g. git merge -s ours) so the higher development branch ignores the cascaded change (the ARSN-600 scenario).

It does not trigger when the PR is already up-to-date with the dev tip: the merge then fast-forwards and no merge commit is created.

Root cause

update_integration_branches() builds acceptable_parents from three git log A..B diffs (prev_set, dst_set, wbranch_set). A..B excludes the boundary commit, so when development/X.Y == development/X.(Y-1), dst_set is empty and the shared dev tip D — the first parent of the integration merge — appears in none of the sets. The check then rejects a parent that legitimately belongs to the target development branch, contradicting the check's own documented intent (its comment explicitly allows parents from "the target development branch").

Fix

Accept a merge parent that is an ancestor of the destination development branch (Branch.includes_commit, i.e. git merge-base --is-ancestor), matching the documented intent.

Genuine rebase / force-push detection is preserved: a rebased source leaves a merge whose offending parent is a feature-branch commit that is not an ancestor of any development branch, so it still raises. The existing BranchHistoryMismatch tests still pass. Two failing-first regression tests cover the non-diverged case:

• test_history_mismatch_nondiverged_dev_branch — manual-resolution path
• test_history_mismatch_nondiverged_dev_auto_integration — auto-created merge of a late PR

Full tox -e tests-noqueue and tox -e tests suites pass (202 tests each).

Jira: PTFE-3343

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.62%. Comparing base (9d05418) to head (e9fda4b).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #271      +/-   ##
==========================================
+ Coverage   89.60%   89.62%   +0.02%     
==========================================
  Files          81       81              
  Lines       10589    10614      +25     
==========================================
+ Hits         9488     9513      +25     
  Misses       1101     1101              

Flag	Coverage Δ
integration	87.23% <100.00%> (+0.03%)	⬆️
tests	87.19% <100.00%> (+0.03%)	⬆️
tests-BuildFailedTest	26.63% <11.53%> (-0.05%)	⬇️
tests-QuickTest	34.16% <11.53%> (-0.07%)	⬇️
tests-RepositoryTests	26.30% <11.53%> (-0.05%)	⬇️
tests-TaskQueueTests	51.35% <15.38%> (-0.12%)	⬇️
tests-TestBertE	65.50% <100.00%> (+0.11%)	⬆️
tests-TestQueueing	53.57% <15.38%> (-0.13%)	⬇️
tests-api-mock	15.33% <0.00%> (-0.04%)	⬇️
tests-noqueue	77.42% <100.00%> (+0.06%)	⬆️
tests-noqueue-BuildFailedTest	26.63% <11.53%> (-0.05%)	⬇️
tests-noqueue-QuickTest	34.16% <11.53%> (-0.07%)	⬇️
tests-noqueue-RepositoryTests	26.30% <11.53%> (-0.05%)	⬇️
tests-noqueue-TaskQueueTests	51.35% <15.38%> (-0.12%)	⬇️
tests-noqueue-TestBertE	61.89% <100.00%> (+0.13%)	⬆️
tests-noqueue-TestQueueing	26.32% <11.53%> (-0.05%)	⬇️
tests-server	28.24% <0.00%> (-0.07%)	⬇️
unittests	43.34% <0.00%> (-0.11%)	⬇️
utests	28.10% <0.00%> (-0.07%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

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

[tcarmet]

Nice work! Please add the extra testcase :)

[benzekrimaha]

LGTM , my comments are not blocking just a nit and a confirmation question