DataArray.chunk() from sparse array produces malformed dask array

[crusaderky]

#3117 by @nvictus introduces support for sparse in plain xarray.
dask already supports it.

Running with:

• xarray git head
• dask 2.2.0
• numpy 1.16.4
• sparse 0.7.0
• NUMPY_EXPERIMENTAL_ARRAY_FUNCTION=1

>>> import numpy, sparse, xarray, dask.array
>>> s = sparse.COO(numpy.array([1, 2]))         
>>> da1 =  dask.array.from_array(s)
>>> da1._meta
<COO: shape=(0,), dtype=int64, nnz=0, fill_value=0>
>>> da1.compute()
<COO: shape=(2,), dtype=int64, nnz=2, fill_value=0>
>>> da2 = xarray.DataArray(s).chunk().data
>>> da2._meta                                                                                                                                                                                                      
array([], dtype=int64)  # Wrong
>>> da2.compute()
RuntimeError: Cannot convert a sparse array to dense automatically. To manually densify, use the todense method.

[shoyer]

I'm pretty sure this is what breaks things:

xarray/xarray/core/variable.py

Lines 939 to 945 in befc72f

	# For now, assume that all arrays that we wrap with dask (including
	# our lazily loaded backend array classes) should use NumPy array
	# operations.
	if LooseVersion(dask.__version__) > "1.2.2":
	kwargs = dict(meta=np.ndarray)
	else:
	kwargs = dict()