Hi Ian, this is expected behaviour because the file exists on the file system and is requested directly through to that without going through the Matrix application.
As Aleks said, as long as you don’t have links on your site pointing to that version of the URL anywhere then it should eventually drop out of SEO, unless of course other people on the internet are linking to it.
Google has to find the URL from somewhere. If the __data path on the site is configured to list files out (not recommended), it might have just crawled the directory structure - otherwise the file had to be linked on that domain from somewhere (not necessarily your site).
Can you review access logs to see if a referrer is listed when the crawler downloads the file?