Add syntax for Supervisor

[armanbilge]

I am quite lazy. This PR adds syntax F[A]#supervise equivalent to Supervisor[F]#supervise(fa: F[A]) which is available whenever an implicit Supervisor[F] is in scope. Not sure where this would go, so I created a syntax package in std.

[vasilmkd]

This looks good to me. Should we port it to CE2 as well?

[vasilmkd]

Unless there are objections to passing Supervisor instances implicitly. Not sure if we have a firm decision on this.

[armanbilge]

Unless there are objections to passing Supervisor instances implicitly. Not sure if we have a firm decision on this.

Discussing on discord, @ChristopherDavenport made this suggestion:

Or if you are in the mood to live dangerously import supervisor.syntax._ in which case if the syntax is used it references that supervisors instance.

Although there may be some trickiness regarding the import. Update: see this example
https://scastie.scala-lang.org/3gjiuAaaRoeEclAUuIcMJg

[djspiewak]

I'm not really a fan of passing Supervisor implicitly, for the same reason I don't think it's a good idea to pass Dispatcher implicitly. Also neither of them are really set up for the "summoning" pattern. With that said, that doesn't really prevent the syntax, it would just mean that the caller would need to pass the supervisor explicitly: fa.supervise(supervisor). Is that worthwhile? It does add to the discoverability, at least in theory.

[vasilmkd]

Personally, I see Dispatcher and Supervisor as similar constructs to CE2's Blocker, which wasn't passed around implicitly. Maybe we should not go through with this change.

[armanbilge]

Ok, sounds like this isn't a great idea after all. I'm happy to adapt my PR for @djspiewak's suggestion fa.supervise(supervisor), otherwise we can close. Thanks for considering it though!

[djspiewak]

Closing this for now; I think the revised syntax is worthwhile, but let's get it into its own PR!