Lattice charts don't work if one group only contains nulls values

[onyxfish]

From wireservice/agate#644

import agate

# Group by county
counties = leso.group_by('county')

# Filter to counties with more than $200,000 in equipment
top_counties = counties.having([
    ('total_cost_sum', agate.Sum('total_cost'))
], lambda t: t['total_cost_sum'] > 200000)

# Group a second time—see the section on "Multidimensional aggregation"
counties_and_items = top_counties.group_by('item_name')

# Aggregate sums by county and item
counts = counties_and_items.aggregate([
    ('quantity_sum', agate.Sum('quantity'))
])

# Sort by item, regroup by county, and then plot a line chart for each county
counts.order_by('quantity_sum', reverse=True).group_by('item_name').bar_chart(label='item_name', value='quantity_sum')

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-104-28220184eafb> in <module>()
     18 
     19 # Sort by item, regroup by county, and then plot a line chart for each county
---> 20 counts.order_by('quantity_sum', reverse=True).group_by('item_name').bar_chart(label='item_name', value='quantity_sum')

/Users/cgroskopf/src/agate/agate/tableset/bar_chart.py in bar_chart(self, label, value, path, width, height)
     29         label_name = self.column_names[label]
     30     else:
---> 31         label_name = label
     32 
     33     if type(value) is int:

/Users/cgroskopf/.virtualenvs/agate/lib/python3.5/site-packages/leather/lattice.py in add_many(self, data, x, y, titles)
    136             title = titles[i] if titles else None
    137 
--> 138             self.add_one(d, x=x, y=y, title=title)
    139 
    140     def to_svg(self, path=None, width=None, height=None):

/Users/cgroskopf/.virtualenvs/agate/lib/python3.5/site-packages/leather/lattice.py in add_one(self, data, x, y, title)
    111             A title to render above this chart.
    112         """
--> 113         series = Series(data, x=x, y=y, name=title)
    114 
    115         for dimension in [X, Y]:

/Users/cgroskopf/.virtualenvs/agate/lib/python3.5/site-packages/leather/series/base.py in __init__(self, data, x, y, name)
     55         self._types = [
     56             self._infer_type(self._keys[X]),
---> 57             self._infer_type(self._keys[Y])
     58         ]
     59 

/Users/cgroskopf/.virtualenvs/agate/lib/python3.5/site-packages/leather/series/base.py in _infer_type(self, key)
     78 
     79         if v is None:
---> 80             raise ValueError('All values in dimension was null.')
     81 
     82         return DataType.infer(v)

ValueError: All values in dimension was null.