Add and test a handler for intervening on random subsets of variables

[eb8680]

The design in #240 and previous iterations includes a number of new components on which the overall correctness of causation and explanation query implementations depends. To reduce the difficulty of reasoning about these implementations and make incremental progress toward landing causation and explanation queries in master, we should pull out these components and specify and test them in isolation.

One of these components is an effect handler that intervenes on randomly selected subsets of a set of possible variables. In the draft design in #240 this handler is called PartOfCause:

class PartOfCause(Protocol[P, S, T]):
    def __init__(self, actions: Mapping[str, Intervention[T]], *, bias: float, prefix: str): ...
    def __call__(self, fn: Callable[P, S]) -> Callable[P, S]: ...
    def __enter__(self): ...
    def __exit__(self, *args): ...

The implementation in #240 is quite compact thanks to #244's undo_split and the syntactic sugar of contextlib.contextmanager:

@contextlib.contextmanager
def PartOfCause(
    actions: Mapping[str, Intervention[T]],
    *,
    bias: float = 0.0,
    prefix: str = "__cause_split_",
):
    preemptions = {
        antecedent: undo_split(antecedents=[antecedent])
        for antecedent in actions.keys()
    }

    with do(actions=actions):
        with BiasedPreemptions(actions=preemptions, bias=bias, prefix=prefix):
            with pyro.poutine.trace() as logging_tr:
                yield logging_tr.trace