Refactor Nucleus and Zome Calls to pure reducers

[lucksus]

• I have added a summary of my changes to the changelog

Purify call reducers

The main thing here is to replace ExecuteZomeFunction (the action and the reducer) by an action creator that returns a future. Just as we did everywhere else except here. This was overdue and happened now.

Clean hdk::call

The side effect is that we can use this action creator, the future return value and our home-grown block_on executor in hdk::call, which hopefully gets rid of the flaky race condition. (and if not, we now have a much better overview of what is really happening)

Steps roughly in commit order

• Break down nucleus/mod.rs that contained all reducer functions into separate files under nucleus/reducer
• Add action creator call_zome_function that spins of a WASM instance without any State change prior and that only dispatches Action::ReturnZomeFunctionResult after the fact
• Remove ExecuteZomeFunction (the action and the reducer)
• Refactor hdk::call to also use above action creator
• Remove Call (the action and the reducer)

SignalZomeFunctionCall

After discussing this in our heartbeat with @maackle, I've readded an action that gets dispatched when a zome function gets called. Since there is no reducer for this action (i.e. no state change) I've named it Signal... It is used by the nodejs test waiter as a signal to attach waiters..

[zippy]

Looking good! I just renabled the call_test_assigned test and it's passing for me

[zippy]

Hope this works to solve the flaky tests!

fixed

[zippy]

@lucksus I ran app-spec locally and it works just fine on my ubuntu system. Not sure why it's failing here. In any case this is much better than before...

[maackle]

I've not had a chance to read this thoroughly, but I see no red flags so I will approve now for timeliness. I'll read through now and can raise issues as followup if need be, but I'm not expecting that to be the case.