Two qubit gates error model can be misleading

[Naphann]

Problem:
The current error model on two-qubit gates can be set independently at each qubit. This can be misleading and unrealistic.
In real devices, the error rate of two-qubit operations depends on each pair of qubits.

What value should be applied when the two qubits have different CNOT/CZ error rates?

Suggested Solution:
Move two-qubit error models to be parameters of QNIC instead of StationaryQubits.

Bonus points:
Add the capability to define error rates between pairs of qubits.