Check diagnostic disclosure, responsive peer controls, and current-connection
document counters after normal bidirectional synchronisation. Verify that
the native API cleanup callback runs once through normal and failed unload.
Retain the existing Fetch, recovery, data-preservation, and reload checks,
and describe the native scenarios in the E2E guide.
Keep API logging callable after display cleanup by forwarding late messages to the console. Abort conflict-check admission and pending dispatch, and propagate cancellation to waiters.
Let started data operations follow their existing success and failure paths without delaying database close. Verify repeated shutdown, instance isolation, and unload/reload in unit, PouchDB, and real Obsidian tests.
vorotamoroz