I am splitting up bug #52919. First the smallest issue, that I called a buglet there:

According to the documentation of sort: Sorting of complex entries is done first by magnitude ('abs (Z)') and for any ties by phase angle ('angle (z)'). However, consider the following code:

a=complex([-5; 0; 5;0;-5],[+0;5; 0;-5;-0])
abs(a)==5
angle(a)
angle(sort(a))

That is, all entries have the same length and thus should be sorted according to angle. But they are not due to inconsistent treating of +0i and -0i on the negative axis. This is due to the comparison on complex values:

a<=5

As Rik noted in comment 14 of #52919, if in the comparison an angle of -pi appears, it is mapped to +pi. So this is a documentation bug for sort and the comparison operators, of course only of low priority.