Spike and Segments datashader operation not renaming dimensions correctly

The following example attempts to datashade Spikes elements, showing one way that works as expected and two that don’t:

# In[1]
import holoviews as hv
import numpy as np
import pandas as pd
from holoviews.operation.datashader import rasterize
hv.extension('bokeh')


# In[2]
arr = np.random.rand(10) * 10
df = pd.DataFrame({'xpos':arr})
rasterize(hv.Spikes(arr), aggregator='count')  # Works


# In[3]
rasterize(hv.Spikes(df.xpos), aggregator='count') # Doesn't work

# In[4]
rasterize(hv.Spikes(df, ['xpos']), aggregator='count') # Also doesn't work

The error message in both cases:

``` WARNING:param.dynamic_operation: Callable raised "DataError("xarray Dataset must define coordinates for all defined kdims, [Dimension('xpos')] coordinates not found.\n\nXArrayInterface expects gridded data, for more information on supported datatypes see http://holoviews.org/user_guide/Gridded_Datasets.html")". Invoked as dynamic_operation(height=400, scale=1.0, width=400, x_range=None, y_range=None) ```

This is somewhat surprising as a numpy array works, but a Series (which is array-like) does not.

[philippjfr]

This has been fixed.

[philippjfr]

This seems like a bug in datashader to me in that the DataArray returned by the line aggregator does not respect the name of the x-column you give it. I can push a simple fix in HoloViews and then file an issue in datashader.