Rectangle.fromCartesianArray and Rectangle.fromCartographicArray don't find the smallest rectangle when positions cross the IDL

In this example `Rectangle.FromCartesianArray` finds a rectangle that spans nearly the entire globe. The blue rectangle is actually the smallest rectangle option from the collection of points.

```js
var viewer = new Cesium.Viewer('cesiumContainer');

var positions = Cesium.Cartesian3.fromDegreesArray([
120, 30,
179, 30,
-179, 20,
-10, 30,
0, 0
]);

for (var i = 0; i < positions.length; i++) {
viewer.entities.add({
position : positions[i],
point : {
pixelSize : 10,
color : Cesium.Color.YELLOW
}
});
}

var redRectangle = viewer.entities.add({
rectangle : {
height: 0,
coordinates : Cesium.Rectangle.fromCartesianArray(positions),
material : Cesium.Color.RED.withAlpha(0.5)
}
});

var expectedRectangle = viewer.entities.add({
rectangle : {
height: 0,
coordinates : Cesium.Rectangle.fromDegrees(-10, 0, -179, 30),
material : Cesium.Color.BLUE.withAlpha(0.5)
}
});
```

Start with the Rectangle.fromCartesianArray and Rectangle.fromCartographicArray entry points and reproduce the supplied positions that cross the IDL. Compare the returned rectangle with the expected smallest rectangle in the example; done means both methods choose the smallest spanning rectangle for this case and the behavior is covered by a regression test.

Written by the indexing model from the issue text.