Fix incorrect coefficient in mixed-state entanglement swapping

[ExistenceOne]

The calculation of ne.d in MixedStateEntanglement.swapping() incorrectly uses self.c * epr.d instead of self.c * epr.b.

For Bell-diagonal states ordered as $(\Phi^+, \Psi^+, \Psi^-, \Phi^-)$, the correct coefficient is:

$$ d_{\mathrm{out}} = a_1d_2 + b_1c_2 + c_1b_2 + d_1a_2. $$

This typo produces incorrect output probabilities. The subsequent normalization can mask the error by rescaling the coefficients.

It can also cause a ZeroDivisionError. For example, swapping inputs with c=1 and b=1 should produce d=1, but the current implementation produces four zero coefficients and fails during normalization.

This change replaces the incorrect term:

- ne.d = self.a*epr.d + self.b*epr.c + self.c*epr.d + self.d*epr.a
+ ne.d = self.a*epr.d + self.b*epr.c + self.c*epr.b + self.d*epr.a