Monitor fast path perf

[kouvel]

Try to improve the monitor fast path code generation to get it to get closer to asm fast path perf by either force-inlining spin loops or moving spin loops to the slow path, such that the slow path call at the end is the only call.

• Depends on https://github.com/dotnet/coreclr/issues/13980

[jkotas]

GetThread is inlined after dotnet/coreclr#14398

[kouvel]

Not likely to happen - nice to have but not necessary, makes sense to do but no data to substantiate