engine/cleanupmgr: move alert log operations to job queue

[mastercactapus]

Description:
The alert log cleanup manually scans the alert_logs table for entries belonging to alerts that no longer exist. Because of the volume, there is no ON DELETE CASCADE trigger (it would make it impossible to delete some services, for example); this is the garbage collection operation to compensate.

This PR moves the alert log cleanup operation to the job queue system.

• The look-for-work component breaks the alert logs table into blocks of 100k, and schedules a job for each block
• Each of these jobs will scan/cleanup 5k rows at a time until complete
• This operation will run once per week, or if restarting GoAlert, up to once per day.
• The priority is set to the lowest of all cleanup operations.