shade function of datashader.transfer_functions does not implement 'span' parameter correctly.

[xuanvinhilb]

Description

If we do not provide the span parameter, the function works correctly. For example:
img = ds.transfer_functions.shade(agg, cmap=customColorMap, how='linear')
However, the function behaves incorrectly when the span parameter is provided. For example:
img = ds.transfer_functions.shade(agg, cmap=customColorMap, how='linear', span = [valMin ,valMax])
This is because the data has been shifted with the offset = min(agg). However, the span value is not shifted accordingly. One way to fix it is to shift the span value

suggested fix:

In the function _interpolate(), we modify the function as follows:
CHANGE FROM:
if span is None:
span = [np.nanmin(data), np.nanmax(data)]
else:
if how == 'eq_hist':
# For eq_hist to work with span, we'll need to store the histogram
# from the data and then apply it to the span argument.
raise ValueError("span is not (yet) valid to use with eq_hist")
span = interpolater(span,0)

TO:
if span is None:
span = [np.nanmin(data), np.nanmax(data)]
else:
if how == 'eq_hist':
# For eq_hist to work with span, we'll need to store the histogram
# from the data and then apply it to the span argument.
raise ValueError("span is not (yet) valid to use with eq_hist")
if 'offset' in locals():
span = interpolater(span - offset,0)
else:
span = interpolater(span,0)

[jbednar]

Duplicate of #368 . Can you try what is suggested there, and if it helps make a PR so that we can review this fix?

[xuanvinhilb]

I read the thread 368 already. I did not check whether it fix my problem. However, what I suggested work for me.