1) Entropy is supposed to be a state function so shouldn't any loop on the PV plane bring the system to its initial state and thus its initial entropy?
Pair of values $P,V$ may not be enough to specify the state of the system, there may be other thermodynamics state variables. If so, it is possible system returns to the same values $P,V$, but it is in a different thermodynamic state because the additional quantities have different value.
But when the pair of values $P,V$ is sufficient to describe thermodynamic state, like it is for a simple homogeneous system such as gas far from condensation, its entropy is function of $P,V$ only and the system indeed gets the same entropy after the values of $P,V$ are restored.
This is true also for any irreversible cycle that has a point where the state is thermodynamic equilibrium state with definite $P,V$. If the the system gets back to such state, it does not matter whether this happened reversibly or not - the entropy returns to the same value characteristic for the equilibrium state $P,V$.
The claim that entropy increases when irreversible cycle is performed means that the entropy of (system + its environment (heat reservoir)) increases. That way, the entropy of the system may return to its original value, while the total entropy still increases.