Async http client

[SgtMarmite]

Things to discuss:

• naming of retry_status_codes x status_forcelist
• retries x max_retries
• tests coverage
• any missing functions?
• are more examples needed?
• fix httpx version or not?
• currently, the backoff time is calculated like backoff = self.backoff_factor ** retry_attempt and I do not know if that is sufficient.

I do not know if I need to change anything in .github/workflows.

TODO:

• Switch https://bitbucket.org/kds_consulting_team/kds-team.ex-chartmogul/pull-requests/3 to this one
  @davidesner @bakobako

[SgtMarmite]

Currently all requests urls are logged into console. This is not intended functionality and I will have to find a way to disable that.

[davidesner]

I think this is not needed and the retry can be handled by the same logic as in the normal client using httpx.Retry and by starting the client(session) with retry parameter see example here

[davidesner]

Is there a way to achieve this without having the main_async function async? E.g. something like this My idea was to have a function, that would run and wait for multiple jobs and just returned all the results (with the disadvantage of having to store all results in memory). Then I would just call something like results = client.run_multiple_sync() in my sync function. Is that possible?

Also I somehow dislike the name process_multiple what does it do? It actually sends multiple requests (not jobs) and returns list of result futures right? The request is basically the parameters of _request_raw http.py equivalent. I think it should be named something like run_multiple_requests(). Also I can imagine having equivalent for each method, e.g. post_multiple

[soustruh]

Closing this PR in favor of #9