BUG: Compressed matrix indexing should return a scalar

[peterbell10]

Fixes gh-10206

This matches how indexing was done prior to 1.3:

scipy/scipy/sparse/compressed.py

Lines 885 to 889 in a905607

	return self.data[start + insert_pos_left:
	start + insert_pos_right].sum(dtype=self.dtype)
	else:
	return np.compress(minor_index == self.indices[start:end],
	self.data[start:end]).sum(dtype=self.dtype)

[tylerjereddy]

Looks like SciPy CI isn't running at the moment! Or at least not showing up in the PR at the time of writing.

It should be possible to write a short regression test for this, right?

[peterbell10]

Added extra check into the existing indexing test. Or would it be better as its own test?

[perimosocordiae]

Looks good, thanks for the fix.

[perimosocordiae]

I think the test is fine currently. Just need to wait for the CI to come back.

[tylerjereddy]

Alright, there's an approving review on the code + test from CJ and CI is all green now so I'll put this in, thanks @peterbell10 & CJ.