Async client using httpx

[tiwilliam]

Please note that this PR include commits from #54 which should be rebased before merge.

This PR adds an asynchronous client backed by httpx. I've decided to keep the sync clients API as is for full backwards-compatibility, it simply wraps the async code.

I've updated documentation accordingly and also took some freedom to restructure parts of it to make more sense with this new client setup.

I also created some VSCode project setting for this project, no strong opinion to include them but took me some minutes to get the environment rolling with linter and formatter without it.

Example usage:

>>> import bankid
>>> client = bankid.AsyncBankIDJSONClient(certificates=(
...     'path/to/certificate.pem',
...     'path/to/key.pem',
... ))
>>> await client.authenticate(end_user_ip='194.168.2.25')
{
    'orderRef': 'ee3421ea-2096-4000-8130-82648efe0927',
    'autoStartToken': 'e8df5c3c-c67b-4a01-bfe5-fefeab760beb',
    'qrStartToken': '01f94e28-857f-4d8a-bf8e-6c5a24466658',
    'qrStartSecret': 'b4214886-3b5b-46ab-bc08-6862fddc0e06'
}

...and sync:

>>> import bankid
>>> client = bankid.BankIDJSONClient(certificates=(
...     'path/to/certificate.pem',
...     'path/to/key.pem',
... ))
>>> client.authenticate(end_user_ip='194.168.2.25')
{
    'orderRef': 'ee3421ea-2096-4000-8130-82648efe0927',
    'autoStartToken': 'e8df5c3c-c67b-4a01-bfe5-fefeab760beb',
    'qrStartToken': '01f94e28-857f-4d8a-bf8e-6c5a24466658',
    'qrStartSecret': 'b4214886-3b5b-46ab-bc08-6862fddc0e06'
}

Preview of documentation updates:

Preview of updated getting started:

[tiwilliam]

Resolves #53.

[Majsvaffla]

I only looked at this briefly for now. Nice things 😎

[Majsvaffla]

Why don't keep the dict literals here? Is it a linter/formatter thing? Not that it matters a lot in this case but AFAIK it performs better with literals.

[tiwilliam]

Only a personal preference, like getting better syntax coloring. No strong opinion though.

[hbldh]

@tiwilliam I would like it merge this into pybankid, because it is functionality that is highly desired, but there are too many things I want to separate in this.

If you develop a async client, I would prefer to have the async client and its documentation added, not also some cleaning of Python 2.7 relics, removal of file headers you did not agree with, a restructuring of the documentation to how you prefer to have it and definitely not replacing the fully functioning synchronous client with a asyncio wrapper around your async client. For synchronous handling, I would prefer to use the regular httpx.Client class instead of wrapping the httpx.AsyncClient in a temporary event loop and run it.

These should in my opinion be done in several PRs, each one focusing on one thing. This is just too much disparate changes in one PR. It changes everything and potentially destabilizes things in doing so.

That said, I plan to merge this into the develop branch and do some reworking of the internals. Thank you for you contribution, but please do smaller changes in the future, focusing on one thing at a time.

[tiwilliam]

@hbldh I could have cleaned this up a bit or split it in separate PRs and iterated on keeping the sync code approach if you just would have given me the chance after feedback. Some of the Python 2.7 changes are from #54 and can be easily rebased as mention in the first comment, I wanted to avoid some merge conflicts which could have been messy.

[hbldh]

Version 1.0.0 of PyBankID, with async and sync clients integrating with the v6 BankID API, has now been officially released and is available on PyPI.