Irit: do you still reproduce the issue using gc.disable()? Or you can try different values to call gc.set_threshold(). You may also try different values in ns.huntrleaks, like: ns.huntrleaks = (3, 20, 'tt_out.py'). With 20 iterations, do you still reproduce the leak?

To make the issue more likely, you can change check_rc_deltas() in Lib/test/libregrtest/refleak.py. Try:

    def check_rc_deltas(deltas):
        return any(delta >= 1 for delta in deltas)

With this change, the script stops immediately:

test leaked [-4, 4, -4] references, sum=-4
test leaked [-4, 4, -4] memory blocks, sum=-4

Using 20 iterations, I get:

test leaked [-4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4] references, sum=0
test leaked [-4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4, -4, 4] memory blocks, sum=0