How to expose/organize `<daemon>d` entryoints?

[goodboy]

As per the reorg in #483 I think it's worth replacing all the daemon entry points currently defined in piker.service._daemon to possibly submods in each sub-system/package.

As an example, for the data layer, we could move maybe_spawn_brokerd() and spawn_brokerd() into a piker.data._daemon (though we probably want to move this particular daemon as datad once we split broker vs. rt feed tasks into separate actors).

The idea being that each sub-sys defines it's own "endpoints" for different services that can be started by name from the .service._mngr.Services API?

I feel like this would make it clear which systems can be run on separate hosts as well as how execution workloads are generally separated logically in our code base?

---

Another thing i've been groking is whether we should move our Qt integration via trio-guestmode in .ui._exec maybe into a .ui._daemon as well since eventually we want to be able to spawn UI actors under pikerd such they are supervised in the same was services and don't require a diff root actor for supervision/respawn?

Main issue for this support in tractor: https://github.com/goodboy/tractor/issues/281