Setup long-term Onionperf instances to support S61 experiments

This initially involves setting up one AWS OP instance managed with Ansible in the HK region, and integrate the data from it in a Collector instance. Instances in other locations may be required in the future. AWS has options to deploy in other zones, including Brazil and Bahrain.

The preferred option is to store the experimental data alongside data from our other long-term instances, but not graph it on metrics.tpo. Something to keep in mind is that any time we make a configuration change on our current Collector we run the chance of missing out on descriptor data if it's not done right (these changes are a manual process).

https://collector.torproject.org/archive/onionperf

https://gitlab.torproject.org/tpo/metrics/onionperf/-/wikis/longterm/data