Only _WIN32 is defined by the compiler, and WIN32 is not standardized
and may be missing.

Closes #169

This eliminates some overhead, because the compiler doesn't need to
consider these functions throwing.

Handles abstract sockets on Linux.

No GLib memory allocation.

No GLib memory allocation.

getnameinfo() doesn't work well - it always returns "localhost".

Replaces GLib's GError.

GLib 2.16 was released more than 4 years ago.  Let's remove some cruft
from the glib_compat.h header, and avoid new cruft to it.

In a C file, that is too late, because _mingw.h defaults to 0x0502,
and ours would be illegal redefintion.

Wrap close(), use closesocket() on WIN32/WinSock.

If a connected host disappears without our knowledge, as can happen over
wireless or a hibernating machine, we continue to hold the port open waiting
for messages. Because we never try to send anything down this now-broken
pipe, the connection will sit idle taking up a slot in our allowed incoming
connections list.

If enough of these happen, an unintended Denial of Service takes place,
where all connection slots are filled with now-broken, never ending
connections. Setting the TCP keepalive option at least allows these to time
out after the default two hours, which is sufficient in the non-malicious
case.
Signed-off-by: Dan McGee <dan@archlinux.org>

In the winsock headers, the setsockopt() argument is declared as
"const char *", not "const void *".

After we've been hit by Large File Support problems several times in
the past week (which only occur on 32 bit platforms, which I don't
have), this is yet another attempt to fix the issue.

Changed the wrappers for pipe(), socket(), accept().  On WIN32, this
does not work for pipe().

Added the "fd_util" library, which attempts to use the new thread-safe
Linux system calls pipe2(), accept4() and the options O_CLOEXEC,
SOCK_CLOEXEC.  Without these, it falls back to FD_CLOEXEC, which is
not thread safe.

This is particularly important for the "pipe" output plugin (and
others, such as JACK/PulseAudio), because we were heavily leaking file
descriptors to child processes.

g_set_error() is a printf-like function, and expects a format string.
Using the return value of gai_strerror() is unsafe.

g_strerror() is more portable, and guarantees that the returned string
is UTF-8 encoded.