Faulty 1D DOK indexing

[sayandip18]

Describe the bug
1D DOK arrays lets the user "access" elements at higher dimensions

To Reproduce

>>> import numpy as np; import sparse
>>> shape=(5,)
>>> density=0.2
>>> s = sparse.random(shape, density, format="dok")
>>> s
<DOK: shape=(5,), dtype=float64, nnz=1, fill_value=0.0>
>>> s[1]
0.0
>>> s[2]
0.0
>>> s[1,1]
<DOK: shape=(2,), dtype=float64, nnz=0, fill_value=0.0>
>>> s[1,1,1]
<DOK: shape=(3,), dtype=float64, nnz=0, fill_value=0.0>
>>> s[1,1,1,1]
<DOK: shape=(4,), dtype=float64, nnz=0, fill_value=0.0>
>>> s[1,1,1,1,1]
<DOK: shape=(5,), dtype=float64, nnz=0, fill_value=0.0>
>>> s[1,1,1,1,1,1]
<DOK: shape=(6,), dtype=float64, nnz=0, fill_value=0.0>
>>> s[3,4,5]
<DOK: shape=(3,), dtype=float64, nnz=1, fill_value=0.0>
# and so on

Expected behavior
Should give IndexingError

System

• OS and version: Ubuntu 20.04
• sparse version: master
• NumPy version: 1.19.2

[hameerabbasi]

Is this on master or on another branch/version of code?

[sayandip18]

This is on master.