Improve separation of encointer-stuff

[clangenb]

I think we have can get better in separating encointer stuff from the rest of the code. The current state is that whenever needed, we just added an encointer module there. However, I think the better approach would be to have the encointer stuff all in one place, and in crates instead of in submodules, which can then be imported in the app-libs crates. In some cases, this is absolutely straightforward, for other cases, we need some better separation of code, so the encointer crate do imports without introducing dependency cycles.

Target structure with the current code I see:

• ./encointer # top level folder (not a crate)
  • stf/ // <- should be straightforward, see Separate trusted calls/getters from rest. #36
    • trusted_call
    • trusted_cetter
  • api-client-extensions // <- should be straighforward
  • cli/ // <- here we first need to extract some utils from the current cli code, so we don't get dependency cycles. Todo: define issue
    • commands
    • trusted-commands

The scheduler STF interface needs some more thinking, in how we can cleanly solve this without having to touch the executor, but it should be doable too. We need to think more generically here so that we can implement in app-libs an: on_update_state function, that the executor will call agnostic to the actual business logic.