Questionable tearing in PowerSysPro leads to initialization failure

[Francesco Casella]

Please check ​PowerSysPro.Examples.OneProdLoadOneLineOneLineOneLoad. The initialization fails because of a division by zero in a torn equation.
The zero term is line2.Y.re, which depends on line2.G = 0.

This choice of parameter is perfectly legitimate, the line is purely resistive and has zero shunt admittance, which should lead to a solvable system of equations. In fact, by setting --tearingMethod=noTearing the system can be initialized successfully.

I guess we should improve the tearing algorithm to avoid this kind of situation. This may actually have a positive influence on other failing models in the testsuite.