[voronoi] issue with pss()

1.76.0 produces this voronoi output:

## 1

```
0
3
673 903 -985 -362
-238 248 -179 453
-136 323 -90 419
```

here`pss([-985,-362],[-179,453,-238,248],[673,903,-985,-362])` finds a CE @[-2674.2,1851.9];


## 2

```
0
3
519 -635 342 -158
-440 707 661 -976
269 -113 507 -171
```

here `pss([519,-635],[519,-635,342,-158],[269,-113,507,-171])` finds a CE @ [788.2,-535.1]


## 3

```
0
3
597 -23 533 65
716 302 455 245
-716 412 999 356
```

here `pss([-716,412],[-716,412,999,356],[533,65,597,-23])` finds a CE @ [-784.19,-1676.38]


## 4

in example ”primary_29.txt" `pss([-1,1], [0,1,0,0], [0,0,-1,1])` finds a CE @ [-0.59,1.41]


In these examples `pss()` solves for CE solutions where the (infinite line) segments tangents the CE, but the segment falls short of reaching that tangent point.