Summing Map with uncertainties over all dimensions fails

[atrettin]

When summing a map that has errors over all of its dimensions with map.sum(), the following error occurs:

---

AttributeError Traceback (most recent call last)
in
----> 1 mu_map.sum()

</afs/ifh.de/group/amanda/scratch/trettin/fridge3/fridge3/lib/python3.6/site-packages/decorator.py:decorator-gen-128> in sum(self, axis, keepdims)

/afs/ifh.de/group/amanda/scratch/trettin/pisa/pisa/core/map.py in new_function(*args, **kwargs)
201 new_state[slot] = deepcopy(getattr(self, slot))
202 if len(new_state['binning']) == 0:
--> 203 return np.asscalar(new_state['hist'])
204 return Map(**new_state)
205 return decorate(original_function, new_function)

<array_function internals> in asscalar(*args, **kwargs)

/afs/ifh.de/group/amanda/scratch/trettin/fridge3/fridge3/lib/python3.6/site-packages/numpy/lib/type_check.py in asscalar(a)
573 24
574 """
--> 575 return a.item()
576
577 #-----------------------------------------------------------------------------

AttributeError: 'AffineScalarFunc' object has no attribute 'item'

This is not a problem when just summing over some axes such as in map.sum(axis=('reco_energy', 'reco_coszen')), and it is also not a problem when the map has no errors. Without errors, a single count over all bins is returned, as one would expect.

I attached a map with uncertainties in JSON for testing:
muon_map.json.zip