fix(coordinator): coalesce parent-child sync notifications to dirty locks

[tykeal]

Summary

Delivers both #683 and #684 as a single releasable unit. Issue #683's description explicitly states "Do not ship this without task-10's quick/debounce scoping" (task-10 being #684), so both are bundled here.

What was already done by #695

PR #695 had already delivered 3 of #683's 5 acceptance criteria:

• override_parent children not mutated/notified (continue guard in _update_child_code_slots; guards in set_pin_on_lock/clear_pin_from_lock)
• Child-unlock parent access-limit decrement notifies the parent coordinator
• Deferred, scoped fan-out (_defer_refresh_listener_updates, _active_refresh_count, _pending_notify_entry_ids, flush on refresh boundary)

This PR implements only the genuinely outstanding parts. Reviewers should not expect a larger diff — the foundation was already in place.

Commits

1. f7b9c18 — fix(coordinator): coalesce parent-child sync notifications to dirty locks

Adds _parent_sync_transaction, an @asynccontextmanager that suppresses per-slot async_schedule_keymaster_notifications fan-out during the _sync_child_locks child loop. Entry IDs are accumulated in a transaction-scoped set[str] and flushed as a single notification on exit. The context manager is re-entrant (depth counter), exception-safe (finally block), and preserves all_entry_ids=True semantics through the deferral.

Design decision (repo owner): Intermediate Synced.ADDING/Synced.DELETING states and sync_op_started_at are still written to the model exactly as before. Only the notification fan-out is suppressed. This deliberately preserves stuck-state recovery (TestSyncPinStuckStateRecovery) and provider-retry logic, which depend on those fields being set before the provider I/O call.

2. b47dc6a — fix(coordinator): scope quick and debounced refresh to per-entry timers

Replaces the global _quick_refresh: bool and single _cancel_quick_refresh/_cancel_debounced_refresh handles with per-entry equivalents:

• _quick_refresh_entry_ids: set[str] — tracks which entries need a follow-up refresh
• _cancel_quick_refresh: dict[str, Callable] — per-entry quick-refresh timer handles
• _cancel_debounced_refresh: dict[str, Callable] — per-entry debounce timer handles

Each set_pin_on_lock/clear_pin_from_lock/_sync_pin/_update_child_code_slots now records only the affected entry ID. Per-entry timers fire through async_refresh_lock(entry_id), which serialises via self._debounced_refresh.async_lock() — the same asyncio lock from the parent DataUpdateCoordinator. This guarantees no independent simultaneous polling is reintroduced.

Shutdown and _delete_lock cancel all/per-entry handles respectively, preventing dangling async_call_later callbacks.

3. e534907 — test(coordinator): tighten per-entry refresh assertions and fix stale test name

Replaces truthiness-only assertions on _quick_refresh_entry_ids (e.g. assert coordinator._quick_refresh_entry_ids) with exact set equality (e.g. == {"child_id"}). This catches wrong-entry-ID bugs that the old assertions would miss. Also renames test_scoped_refresh_keeps_shared_debounce_for_other_dirty_locks → test_scoped_refresh_preserves_other_entry_debounce_timer to reflect the per-entry design.

Validation

• pytest tests/: 1092 passed, 1 deselected
• ruff check custom_components/ tests/: All checks passed
• ruff format --check custom_components/ tests/: All formatted
• mypy custom_components/keymaster/: No issues found

Mutation testing

All new tests were mutation-verified. Key mutations validated:

• Removing the transaction check from async_schedule_keymaster_notifications
• Removing the finally depth decrement
• Removing the coalesced flush at transaction exit
• Recording the parent entry ID instead of the child's at _update_child_code_slots
• Skipping individual cancel() calls during shutdown
• Cancelling ALL handles when only one entry's should be cancelled

Closes #683
Closes #684

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.25%. Comparing base (cdb4922) to head (17e1ba6).
⚠️ Report is 230 commits behind head on main.
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@             Coverage Diff             @@
##             main     #707       +/-   ##
===========================================
+ Coverage   84.14%   94.25%   +10.10%     
===========================================
  Files          10       42       +32     
  Lines         801     5480     +4679     
  Branches        0       30       +30     
===========================================
+ Hits          674     5165     +4491     
- Misses        127      315      +188     

Flag	Coverage Δ
python	94.15% <100.00%> (?)

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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated no new comments.