Document the many distributed "modes"

[goodboy]

The runtime has been built to handle distributed, many host, deployments
from the ground up; there are quite a few configurations for this (and
many more to come B)

---

Let's start listing the possible ploys..

• full pikerd actor tree runs on local host but (ib) API gateway is on
  remote host and (maybe) the tsdb is running on a remote (in docker).

• pikerd runs on local host

  • +remote brokerd + samplerd
  • +remote emsd
  • +remote fsp cluster

• pikerd runs remote but "some" of it's subactors are spawned
  locally:

  • brokerd is local
  • rest of tree is in pikerd tree

.. more to come

---

Doc all the configs

• $HOME/.config/piker/conf.toml
  • [network] section for laying out bind and connect sockaddrs
  • some some drafted out ideas in another file somewhere..