`allows-inexact-div-by-zero`

Dividing by zero can sensibly yield either some floating point representation of an infinity or a NaN, or it can signal a division by zero error. The choice of what to do can sensibly depend on whether the division is being carried out exactly and, separately, whether the zero being divided by is exact. Compare allows-div-by-inexact-zero and allows-div-by-exact-zero.