Drop sync

[lovelydinosaur]

Trawling through and dropping sync support everywhere.

There's a bunch of stuff that I think we really need to slim down on in order to focus on a really tight 1.0 release. I also don't think we've got a good approach onto sync support as things stand.

Subclassing in order to provide sync+async, leaading to BaseRequest, BaseResponse, AsyncRequest, AsyncResponse, Request, Response, Client, AsyncClient, Dispatcher, AsyncDispatcher just doesn't feel legible when getting to grips with any particular part of the codebase.

I think we need to get the asyncio+trio support absolutely squared away, and have any sync support come in a 2.0 version, that also keeps any bridging more clearly isolated.

[florimondmanca]

Just skimmed through this for now. Generally I agree with the change.

I see test_client.py and async_test_client.py both still exist — do we want to merge these in one test_client.py file? And if so, do we want that to happen in this PR? I’d be fine with combining in a separate PR as there might be overlapping test cases that wouldn’t be quite as easy to review.

[florimondmanca]

Interesting diff on this class, right? Related to the codegen approach folks from urllib3-async pointed out :)

[florimondmanca]

Also, refs #522.

[florimondmanca]

Here's a more thorough review!

[lovelydinosaur]

I think this should be ~up to scratch now.

Big jump to make, and likely to be unpopular in places, but I think we've got more than enough on our plate to fully square away before coming back to a decision on best tack for bridging-to-sync.

[florimondmanca]

Let’s do this. 🎢

[lovelydinosaur]

[florimondmanca]

Boom 💥