[no squash] porting.h maintenance

[sfan5]

To do

This PR is Ready for Review.

How to test

no need really

[srifqi]

Using self-compiled MinGW build on Windows 10,

fps_max	master	This PR
30 FPS	30±1 FPS	30±1 FPS
60 FPS	61±1 FPS	58±1 FPS
120 FPS	122±1 FPS	115±1 FPS
240 FPS	244±2 FPS	220±1 FPS
(trend)	(higher)	(lower)

In some cases, the Minetest's FPS counter is lower (by around 1 FPS) than the Xbox Game Bar's FPS info using this PR version.

[sfan5]

Hm.
I can reproduce that it undershoots the target FPS in wine (if I trust our FPS counter), this would point towards a bug in the timer code.

[sfan5]

Some debugging code to add to myusleep:

	auto start = getTimeUs();
	WaitForSingleObject(timer, INFINITE);
	auto end = getTimeUs();

	if (fabs(fabs(end-start) - usecs) > 100) {
		char buf[100];
		snprintf(buf, 100, "wanted to sleep %dus actually %dus", (int)usecs, (int)fabs(end-start));
		actionstream << buf << std::endl;
	}
	CloseHandle(timer);

I sometimes see it sleeping up to double the requested time for small amounts like 1000.
Inserting similar code into the Linux code path it never triggers for me.

It's weird because CreateWaitableTimerExW + CREATE_WAITABLE_TIMER_HIGH_RESOLUTION is supposed to be more accurate way to sleep on win32.

[sfan5]

Well I've run out of things to try so I'll convert this PR to only the maintenance changes.

[srifqi]

The code looks good to me. I can not test on other than Windows, though.