Upgrade to cats-effect, fs2 3.0.0-M3 builds

[Daenyth]

This adds some minor api cleanup as well. It shouldn't affect anyone downstream too badly, I think

[Daenyth]

I think at this point the main unresolved question is whether CE3 will recommend using Dispatcher as an implicit. Currently it's not really set up for that, but it makes this code clearer to do so. But if the library instructs people to pass it explicitly, I'll want to change the signatures or do the api change like I was thinking earlier

[Daenyth]

recommendation from CE team is to create locally, it's cheap

https://gitter.im/typelevel/cats-effect-dev?at=5fbbc54afe857d71dc8dad83

Edit: that won't work, as I don't think changing the signature to return F[akka.streams.Source] etc is acceptable. So we will have to take it as an argument

[kubukoz]

WDYT about wrapping all ops in a trait and creating a dispatcher as part of that? Then all the methods inside could hopefully return raw streams/sinks (with no constraints, which is a good thing anyway)?

[Daenyth]

I think it's reasonable, but I wasn't sure. Some thoughts in #84

[Swoorup]

Anything that is blocking this currently? I have been using @Daenyth code is isolation. Seems to do solve my issue since I am using CE3.

[Daenyth]

I haven't had time to work on this lately, and I'm no longer using it at work.

I'd say this is probably "good enough" to push just to have something out, but I don't have bandwidth to roll it out