Following scenario:

• user types into a file and quits without saving
• the file changes on disk from another program without Code running
• user starts again and finds the file dirty as he was leaving it
• user saves and everything is fine

=> this basically bypasses our check to prevent dirty writes to files that have changed meanwhile
=> maybe the backup should store the etag of the file (mtime + size) as part of the backup in the meta data section of the backup (where we should also store the file path) so that this check is not bypassed.