improving the complex number division

[Robert Dewar]

It is indeed surprising to find this entirely naive and quite incorrect
implementation of complex division in gcc. This is certainly a computation
for which absolutely standard algorithms exist. I am not clear whether
Jerome is quoting one of these standard algorithms (in which case a reference
would be helpful), or inventing from scratch (in which case I would still be
a little skeptical, though to my not-really-expert-enough eye, the proposed
replacement algorithm looks correct -- I just don't know if it is the best
choice.