Research how to utilize Sentry's Performance to diagnose performance issues

[jpopelka]

Even after several investigations, we're still seeing intermittent SLO1 performance issues (#1954). So far the identifications of the cause(s) have mostly been guesswork.

Luckily, we've been collecting performance data in Sentry since half of May. They trace the transactions through the whole stack and should show us the bottlenecks more precisely.

For example this page shows the slowest task.steve_jobs.process_message transactions.

• [ ] Study available resources (docs, videos) about how to utilize the Performance
• [ ] Create a small demo for the team about what you've learned
• [ ] Try to identify the causes of SLO1 performance issues
• [ ] Update #1954

[lbarcziova]

This is currently low prio, but we could have a look while working on #1954.