Added operation_kwargs Histogram property

[jonmmease]

Overview

For Histogram elements created using the holoviews.operation.histogram operation, this new property holds a dict of kwargs that can be used with the operation to recreate the Histogram from the same data set.

Or it will contain None, if the Histogram was created directly using the Histogram constructor.

Motivation

Having this information available makes it possible to create histograms with a subset of the original dataset that share the same properties as an original Histogram. For example:

import numpy as np
import holoviews as hv
from holoviews import opts
hv.extension('plotly')

np.random.seed(1)
points = hv.Points(np.random.randn(100,2))

hist1 = hv.operation.histogram(points, cumulative=True, num_bins=10, dynamic=False, normed=False)
hist1.operation_kwargs

{'cumulative': True,
 'dynamic': False,
 'normed': False,
 'bins': [-1.857981864446752,
  -1.419351107321437,
  -0.980720350196122,
  -0.542089593070807,
  -0.10345883594549199,
  0.335171921179823,
  0.7738026783051382,
  1.212433435430453,
  1.651064192555768,
  2.0896949496810833,
  2.528325706806398]}

hist1

hist2 = hv.operation.histogram(points.select(y=(0, None), x=(0, None)), **hist1.operation_kwargs)
hist1 * hist2

This will be especially convenient when combined with #3919, because the original dataset will be available as the hist1.dataset property.

TODO:

• Presever operation_kwargs through clone.

[jbednar]

Sounds good!

[jonmmease]

I made operation_kwargs a private property of the Histogram element so that we're not committing to a particular API at this point. I opened #3941 to discuss what we might want to do with the API in the future.

Merging!