set_loop_type supports reverting back to synchronous code

[pashazz]

Previously r.set_loop_type used a global variable to determine loop type, thus making running asynchronous and synchronous code alongside each other impossible.

Now you can do:

r = rethinkdb.RethinkDB()

To set loop type:

r.set_loop_type('asyncio')

To unset loop type:

r.set_loop_type(None)

This is breaking change: here you should do r.set_loop_type for every RethinkDB object used in asynchronous code. This is probably OK since you'd never use the same RethinkDB object in both synchronous and asynchronous code

[asakatida]

This is abusing __all__ because of the setattr happening in RethinkDB init. Otherwise this is going to override the connect method defined in the class on instance instantiations. Without connect here Python rules implicitly say it's a private method, then we expect it above. I'm not a fan of the confusion in all that. Renaming to make_connection or changing it to a class would solve this.

[asakatida]

@gabor-boros the Codacy issues were all there before, just moved files. I say we just ignore that here. I would like a second set of eyes on my two code comments.

[pashazz]

so everything's ok then?

[gabor-boros]

@grandquista Hell. Unfortunately, I resolved the import related comment but I totally agree, it should be at the top. And I also agree with the renaming what you mentioned.

[gabor-boros]

@grandquista Can we merge this? What do you think?

[asakatida]

@gabor-boros I rolled these changes with my comments in #86 so merge that one and then close this

_{Sent with GitHawk}