syz-cluster: current backlog

[a-nogikh]

#5620 is the first (not yet complete) version of the syz-cluster system, many more changes still need to be done.

Global:

• syzkaller: switch to -mod=mod? #5645 to address the only growing vendor folder
• Storing binary blobs in a separate blob storage leads to synchronization problems: if we don't do two-phase commits, we risk ending up with orphaned files in the storage.
• Optimize pkg/ dependencies: in many cases there's really no need for syz-cluster to use vm/ or sys/, but the transitive dependencies demand copying them into Dockerfiles.
• Adjust to GKE. Set up CI/CD. Formulate the update rollout strategy.
• Consider using mockery instead of workflow.MockService.
• Consider unifying ID and ExtID of Series.

TODO (general):

• Add a boot test.
• Integrate with syz-diff: syz-cluster: assorted improvements #5728
• Set up email reporting.

TODO (series-tracker):

• Collect and monitor the To/Cc list.

TODO (controller):

• Don't schedule a workflow if there's a newer version of the same series.
• Add request field validation.
• When the session is finished, scan for "Running" tests and mark them failed.

TODO (syzbot dashboard):

• Return lists of bug titles found in the last X days.