vector runtime: Incorrect value when dividing by zero using floats

The correct result from sequential runtime:
```
$ echo '{a:2.,b:0.}' | super -c 'values a/b' -
error("divide by zero")
```

The incorrect result from vector runtime:
```
$ echo '{a:2.,b:0.}' | SUPER_VAM=1 super -c 'values a/b' -
+Inf
```