Make inv(::ComplexF64) return zero when input has Inf magnitude

Author: giordano

base/complex.jl

@@ -433,6 +433,7 @@ function inv(w::ComplexF64)
     half = 0.5
     two = 2.0
     cd = max(abs(c), abs(d))
+    (isinf(c) || isinf(d)) && return complex(copysign(0.0, c), flipsign(-0.0, d))
     ov = floatmax(c)
     un = floatmin(c)
     ϵ = eps(Float64)