Setting an image to [0, 0, ...] causes crash

[haesleinhuepf]

Hi @Who8MyLunch ,

first of all, thanks to programming this amazing widget! I love it. I made a library based on it and use it often to explore image data from within jupyter notebooks.

I have a little issue when setting an image to an array of zeros. This is a minimal working example:

import numpy as np
import numpy_image_widget as niw

image = np.asarray([[1,0], [2,3]])

view = niw.NumpyImage(image)

Until here everything is good. If I then set the data:

view.data = np.asarray([[0,0], [0,0]])

I receive this error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
~\AppData\Local\Temp/ipykernel_13780/2881574154.py in <module>
----> 1 view.data = np.asarray([[0,0], [0,0]])

~\miniconda3\envs\STRESS\lib\site-packages\numpy_image_widget\numpy_image_widget.py in data(self, new_data)
     70 
     71         self._data = new_data
---> 72         self.value = ia.compress(self._data, self.format, quality=self.quality)
     73 
     74     @property

~\miniconda3\envs\STRESS\lib\site-packages\image_attendant\compression.py in compress(data, fmt, **kwargs)
     22     """
     23     # Enforce 8-byte data
---> 24     data = utility.setup_uint8(data)
     25 
     26     fmt = fmt.lower()

~\miniconda3\envs\STRESS\lib\site-packages\image_attendant\utility.py in setup_uint8(data, lohi)
     77         lo, hi = lohi
     78         if lo == hi:
---> 79             raise ValueError('Invalid data range: {}, {}'.format(lo, hi))
     80 
     81         data = (data - lo) / (hi - lo)

ValueError: Invalid data range: 0.0, 0.0

It's obviously a division by zero because min and max of the image are equal. If you're busy, I'm happy to send a PR. Just let me know.

Thanks!

Best,
Robert