Release v1.19.1

Release v1.19.1: restore progress heartbeats during hashing, extraction and apply.

Changelog

Fixed

• A very large restore still running could be judged abandoned by restore.stale_after_seconds (default 2 × process_timeout), because archive hashing, tar extraction and file apply wrote no progress: a second restore could then start on top of it. These steps now write progress heartbeats (refresh updated_at) at most every 5 s, or every third of stale_after_seconds when that is smaller, so staleness is measured from the last heartbeat. BackupArchiver::verifyIntegrity() / extractTo() accept an optional heartbeat callable (archive hashed in 8 MiB chunks, tar polled while running); RestoreOrchestrator accepts an optional clock closure.

Full Changelog: v1.19.0...v1.19.1