Multiple db errors while setting auto-tags

[ice-phoenix]

Original report by Mikhail Belyaev (Bitbucket: belyaev.mikhail, GitHub: belyaev-mikhail).

---

Several log messages cause races between setting and resetting of tags on the same submission. These are guaranteed not to cause any inconsistencies (or are they?), but are very annoying.

A typical scenario is as follows:

• Log 1 received (success)
• Log 2 received (success)
• Log 3 received (success)
• Log 1 handler starts dbFind, gets records A, B, C
• Log 2 handler starts dbFind, gets records A, B, C (duh!)
• Log 3 handler starts dbFind, gets records A. B, C
• Log 1 handler deletes records A, B, C, creates record A1
• Log 2 handler deletes record A (knows nothing about A1), which is already deleted, crashtrace
• Log 3 handler deletes record A (knows nothing about A1), which is already deleted, crashtrace