Named steps in Chain

[kurmukovai]

Allow named steps, via OrderedDict, in the Chain, similarly to named steps in pytorch.nn.Sequential:

pipe = Chain(OrderedDIct(
    ('lung ct data', dataset),
    ('filter spacing', Filter(lambda spacing: spacing[-1] <= 2.5, verbose=True),),
    ('select ids', Filter.keep(dataset.ids[:20])),
    ('zoom', Apply(image=lambda x: zoom_to_shape(x, (64,64,64))))
)

This allows to get an access to different preprocessing steps by a name, rather then by the index:

pipe['filter spacing']

instead of:

pipe[1]

to get all the preprocessing up to spacing filtering.