Implement process create time about psutil HOT 13 CLOSED

Implemented on Windows as r232 .

Implemented on Linux as r236 .

A debate to decide what return for PID 0 is also needed.
I modeled Windows and Linux implementations to provisionally return 0.0 (year 1970)
for now.
Returning the system uptime could make more sense.

PID 0 is the kernel, so system uptime would make sense.

Modified special case in create_time to return system uptime for PID 0 on Windows and
Linux as r257 .

Just do document this somewhere aside from the SVN log, in order to return system
uptime as a number of seconds since the epoch we had quite the juggling match on
Python 2.6 due to this issue: http://bugs.python.org/issue3308 Apparently the time.h functions aren't able to be loaded properly when using MinGW
with Python 2.6 on Windows. To get around this I've implemented the calculation of
current time in seconds since the epoch using the Win32 API to get the current time
as a FILETIME struct and converting the value. This is all implemented in
get_system_uptime() using some code I found here: 
http://johnstewien.spaces.live.com/blog/cns!E6885DB5CEBABBC8!831.entry Tested the code and it's returning the correct value (same as time.time() on Windows)
for current time which we can then subtract system uptime seconds from in order to
get system uptime in epoch time.

Note that Python 2.5 works just fine with MinGW time.h functions due to different
version of the visual C++ libs etc. but for compatibility reasons we'll use this
approach to avoid any issues in various environments.

I created issue 39 to discuss about pid 0 special case implementation separately.

Implemented on OS X as r269 .

Implemented on FreeBSD as r272

Updated csets after the SVN -> Mercurial migration: r232 == revision 
7dc906066c68 r236 == revision 4e4234ba4c30 r257 == revision 442aee762cf4 r269 == revision 0c8ab44644c0 r272 == revision 56a85a5d48e9