Removed a bug in QuantumCircuit.toffoli

[Biswayan375]

When the QuantumCircuit.toffoli function is called inside a loop and with a variable array for control variables, the bug is spotted. The scenario is like

n_qudits = 3
qc = QuantumCircuit(qregs=[4 for i in range(n_qudits)])
controls, target = [], n_qudits-1
for i in range(n_qudits-1):
    controls.append(i)
    qc.toffoli((controls, target), 1)
qc.measure_all()
print(qc)

The output should be

(4)|0>--*----*----M--
(4)|0>--|----*----M--
(4)|0>-X(1)-X(1)--M--

depth: 2, width: 3

Rather I got

(4)|0>--*----*----M--
(4)|0>--*----*----M--
(4)|0>-X(1)-X(1)--M--

depth: 2, width: 3

The problem was not with loops, it was with using a variable array for the controls. First I thought the issue was with the QuantumCircuit.__str__ I recently wrote. But this function was dependent on the operator flow itself. So, I found the problem with the parameter qreg, inside which the controls array is. I was passing the array and inside QuantumCircuit.toffoli, it is directly used instead of its copy. So, whenever I change (append a new value) in the array, it gets updated in the Toffoli object as well. I just made it use a copy instead and the problem is solved for now. I have not studied the entire codebase so I don't know how many places it needs to use a copy instead of the original object. It is solved now for the Toffoli only.

I reproduced the issue to get the bug with the following

class A:
    def __init__(self, x: tuple[list, int]):
        self.x = x
l1 = ([0, 1, 2], 2)
a = A(l1)
l1[0].append("bummer")
print(a.x)

which outputs

([0, 1, 2, 'bummer'], 2)