asformat('{csr,csc}') doesn't work on DOK and COO

[sayandip18]

Describe the bug
asformat() does not support conversion of COO and DOK to csr or csc although the same is supported for GCXS.

To Reproduce

>>> import sparse; s=sparse.random((5,5))
>>> s
<COO: shape=(5, 5), dtype=float64, nnz=0, fill_value=0.0>
>>> s.asformat('csc')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/sayandip/sparse/sparse/_coo/core.py", line 1452, in asformat
    raise NotImplementedError("The given format is not supported.")
NotImplementedError: The given format is not supported.
>>> 
>>> s=sparse.random((5,5), format='dok')
>>> s.asformat('csc')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/sayandip/sparse/sparse/_dok.py", line 524, in asformat
    raise NotImplementedError("The given format is not supported.")
NotImplementedError: The given format is not supported.
>>> 
>>> s=sparse.random((5,5), format='gcxs')
>>> s
<GCXS: shape=(5, 5), dtype=float64, nnz=0, fill_value=0.0, compressed_axes=(0,)>
>>> s.asformat('csc')
<CSC: shape=(5, 5), dtype=float64, nnz=0, fill_value=0.0>

Expected behavior
A clear and concise description of what you expected to happen.

System

• OS: Ubuntu 20.04.1
• sparse version: master

Additional context
tocsr and tocsc work as expected for COO object.

[hameerabbasi]

Beware that in fixing this I took away the ability to add compressed_axis=something everywhere, just passing through the kwargs. This usually means compressed_axes can only be supplied if format was GCXS.

[hameerabbasi]

This might conflict a bit with your Hypothesis PR, @sayandip18 but if you follow the small test changes I made, it should be okay.