Fix signed overflow in atan2

As a component of atan2(y, x), the case of x == 1.0 is farmed out to
atan(y). The current implementation of this comparison is vulnerable
to signed integer underflow (that is, undefined behavior), and it's
performed in a somewhat more complicated way than it need be. Change
it to not be quite so cute, rather directly comparing the high/low
bits of x to the specific IEEE-754 bit pattern that encodes 1.0.

ok martin@ pgoyette@ maya@
obtained from FreeBSD

branches: 1.12.94;
__STDC__ is always defined on NetBSD.

More trailing white space.

Fix a brain error from luke while he WARNSifyed libm.
Erik Bertelsen found the missplaced brakes.

branches: 1.9.2;
rcsid facism, WARNSify

Id -> NetBSD

LIBM_SCCS must be defined for the RCS ID to be defined.
This library will soon be used by DJGPP. And since memory is often tight
on DOS machines, there is little benifit to be had from including RCS IDs.

Change int -> int32_t and unsigned int -> u_int32_t to improve portabilty
to machines where int != 32 bits.

Float versions of math functions. From Ian Taylor (ian@cygnus.com), with
minor changes by me.

Don't determine byte order at run time.
This also "solves" the problem of GCC optimization leading to incorrect
results.

Add RCS ID's.

Include <math.h> instead of "fdlibm.h"
Byteorder fix for s_frexp.c.

branches: 1.1.1;
Initial revision