Run usage data calculations in separate queue

[lbarcziova]

As we discussed with the team in our architecture meeting, calculations of the usage statistics are resource-demanding and can block multiple workers (even all), mainly on startup (redeployment), when no data is cached yet. We should avoid this, most probably by having a separate queue where those tasks would be run.

Other alternatives that can be considered:

• utilising task priorities
• cronjobs

When discussing this issue, we also touched on the topic of the data cleanup that might be needed, we should think about that as well.