kernel: convert GlobalSchedulerContext, KAddressArbiter, KScopedSchedulerLockAndSleep, KThreadQueue to new style

At least on Nvidia, glClientWaitSync with a timeout of 0 (non-blocking) is faster than glGetSynciv of GL_SYNC_STATUS.

SpeedLimiting is SC only. Since MC is performance oriented we should check for it first to skip checking use_speed_limit.

And also demote Audren and CoreTiming to High thread priority.

It was experimentally determined to be sufficient.

We want to synchronize RDTSC to real time.

The RDTSC frequency reported by CPUID is not accurate to its true frequency.
We will spawn a separate thread to calculate the true RDTSC frequency after a measurement period of 30 seconds has elapsed.