lower and upper bound for the real part of the slack voltage

[100110110101]

As discussed in #123 I took a closer look at the opf-optimod.
In doing so, a very small suggestion for improvement came to my mind, which I would like to recommend here (it's not a bug):

The lower and upper bound for the imaginary part of the slack voltage are set to zero in line 453 of grbformulator_ac.py. This is clear and reasonable.
But the upper bound of the real part of the slack voltage is set to bus.Vmax and the lower bound is set to -bus.Vmax in lines 444 & 445. I think a tighter formulation is possible without loss of generality: Analogous to setting the voltage angle at the slack bus to zero in polar coordinates, the real part of the slack voltage can be limited between bus.Vmin <= fvar[bus] <= bus.Vmax. This prevents a "mirrored solution" (--> symmetry during optimization) on the negative side of the real axis, which corresponds to an voltage angle of 180°. Therefore, I suggest to add the following code before line 447:
if j == alldata["refbus"]: lbound = bus.Vmin # Modified slack lower bound to prevent a mirrored solution
Furthermore, the quadratic voltage constraint for the slack bus can be omitted.

I assume that this will only bring minor improvements, but I wanted to share the suggestion anyway and put it up for discussion.