Use asyncio for provisioning

• Use asyncio for provisioning. Concurrency is achieved using TaskGroups which automatically collects and raises errors. Any pending tasks are cancelled.
• Use httpx instead of requests
  • httpx has slightly different errors, so we need slightly different error handling
  • use pytest-httpx instead of responses. The provided httpx_mock fixture has similar but slightly different api. By default all mocked requests must be called exactly once. The is_optional and is_reusable parameters can be used to tweak this
• Circumvents a bug in Grafana where a service account may exist, but does not show up in search
• Number of concurrent connections to Grafana are limited, both for performance reasons and to not have the database choke on too many connections from Grafana

changed the description

added 1 commit

• c3ca8e0f - paralellize requests to inprov and reporting provider

Compare with previous version

changed the description

not a lot of performance wins but more complexity.. closing as discussed

closed