Complex Noise section signal power 2?

[mindentropy]

In the section https://pysdr.org/content/noise.html#complex-noise it is mentioned the average power of a zero mean signal using:

power = np.var(x)

In the subsequent line it is mentioned the power of signal n is 2 where n is as follows:

n = np.random.randn() + 1j * np.random.randn()

When I run np.var(n) , the result is 0. So how is the power of signal 2?

[777arc]

try np.var(np.random.randn(100000) + 1j * np.random.randn(100000))

[mindentropy]

@777arc : I get ~2.00. So would it be for sufficiently large N?

For just randn(N) with a large N, I get variance= ~1 which matches mean=0 and variance=1 parameters for randn().
So please correct my understanding, for complex numbers is the variance of real and imaginary parts taken separately and then added to get variance = ~2 ?

Also with relation to the above, I am confused as to why we divide by $\sqrt{2}$ for unity power. Could you please better explain that part?