🔀 REFACTOR: workchain subprocess queing

[chrisjsewell]

A proposal to change how to add a subprocess in a workchain step
(sidenote: this PR is fully back-compatible)

On the user side, basically I think we should replace the use of self.submit/self.to_context/ToContext, with simply self.queue_subprocess.
This better describes what we are actually intending the user to do:

define a (sub/child)process, which will be run until termination, before the next step of the workchain commences, and whose resultant node/outputs will be available to the user in the next step, via a key on the workchain ctx

e.g. change this:

            def step(self):
                self.to_context(child1=append_(self.submit(SimpleWc, val=Int(1))))
                return ToContext(**{'nested.key.child2': self.submit(SimpleWc, val=Int(1))})

to this:

            def step(self):
                self.queue_subprocess("child1", SimpleWc, {"val": Int(1)}, ctx_append=True)
                self.queue_subprocess("nested.key.child2", SimpleWc, {"val": Int(1)})

On the internal code side:

1. renamed some things to not use the term awaitable, which has a very definite meaning in python/plumpy (https://docs.python.org/3/library/collections.abc.html#collections.abc.Awaitable, https://docs.python.org/3/library/typing.html#typing.Awaitable) and instead refer to "process references"

2. Use a dataclass instead of an AttributeDict, since this has various issues for type safety

---

Notes:

• This currently is fully backward-compatible; it doesn't actually remove any of self.submit/self.to_context/ToContext.
  deprecation warnings could be added though.

• Currently self.queue_subprocess works exactly the same as if you did self.submit then self.to_context.
  I don't think this technically correct, though, since I think subprocess submission should only really happen after all the code in the step has been run. Otherwise, you can submit a process, then except the step, and now you have a subprocess running that you don't actually want.