What type of PR is this?

/kind feature

What does this PR do / why do we need it:
The PR uses COOTensor to replace the orginal SparseTensor, and enables the return value type to be COOTensor instead of tuple.
Note:
If users still use SparseTensor to create COOTensor, a warning will appear and remind users to use COOTensor instead, and the SparseTensor interface will not be supported in the near future. For example:

from mindspore import SparseTensor, Tensor, ms_function

indices = Tensor([[0, 1], [1, 2]])
values = Tensor([1, 2.])
shape = (3, 4)

def test_pynative():
    return SparseTensor(indices, values, shape)
test_graph = ms_function(test_pynative)

out1 = test_pynative()
out2 = test_graph()
print(out1)
print(out2)

Output:

Which issue(s) this PR fixes:

Fixes #

Special notes for your reviewers: