Bug: s3_scan_object re-tagged clean object with in_progress

[patheard]

Summary

We recently saw a case in Notify where a scan timed out because a file that was scanned and tagged as clean was then retagged as in_progress before Notify could register the scan result.

The timeline was:

• 15:09: this S3 object was created and a scan was triggered
• 15:09: scan completes with status clean
• 15:09: S3 object is tagged with status clean
• 15:13: S3 object gets tagged with status in_progress

It is not clear why the initial in_progress tag request completed so much later than the clean tag request.