Group delay calculation (phase wrap -180 to 180)

Hi

I'm calculating group delay for branch line coupler port 3 and I get this large peak at center frequency. I'm using ngspice group_delay function. It is clear that its caused by phase transition from -180 to 180 as it is at same frequency as the peak.

Is that in fact correct?

From my own thought process the phase difference is smooth and there shouldn't be any discontinuity in group delay, but I saw otherwise in other post. Also if I calculate angle between vectors -1+j0.001 and -1-j001 its very small.