refactor: ensure experimentalists have a consistent interface

[hollandjg]

... which can accept sequences of tuples, or arrays, or recarrays.

We encounter the problem that different theorists/experimentalists expect and produce different data formats (e.g., numpy versus pandas) and with potentially different dimensions (e.g., 1D vs. 2D numpy array). We should consider standardizing the inputs & outputs, potentially using a pandas data frame.

Perhaps put the functions from #229 into a function decorator which would look like:

@pipewrapper(needs="array")
def sampler(input: np.array):
    output = ... do something with the numpy array ...
    return output

@pipewrapper(needs="recarray")
def sampler_with_recarray(input: np.recarray):
    output = ... do something with the numpy recarray ...
    return output

@pipewrapper(needs="sequence")
def sampler_with_sequence(input: Sequence[ExperimentalConditions]):
    for i in input:
        if i ... fulfils some condition:
            yield i