kernel/scheduler: Make invoke_scheduler atomic to skip unnecessary ICIs.

Following upon a KDL reported by atomozero, on wait-for-free-SMP-messages
on an unblock occurring with interrupts disabled.

In changing cpu_ent, move disabled to the beginning to avoid
enlarging the structure unncessarily.

On a 4-core VM, this skips around 200 sends during boot, and
over 3000 during a rebuild of HaikuDepot + mime_db (cold). On
bare metal (i3, 2x2), it skips around 150 during boot, and a bit
below 3000 during a rebuild of HaikuDepot + mime_db (over a much
longer time than in the VM, as the hardware is slower.)

Performance in the VM doesn't look much different. But this might
help in VirtualBox, or other situations where ICI latency is far
above what it should be.

headers/private/kernel/cpu.h

@@ -50,8 +50,9 @@ typedef struct cpu_topology_node {
 
 typedef struct CACHE_LINE_ALIGN cpu_ent {
 	int				cpu_num;
+	bool			disabled;
 
-	// thread.c: used to force a reschedule at quantum expiration time
+	// used to force a reschedule at quantum expiration time
 	bool			preempted;
 	timer			quantum_timer;
 
@@ -72,8 +73,7 @@ typedef struct CACHE_LINE_ALIGN cpu_ent {
 
 	Thread*			running_thread;
 	Thread*			previous_thread;
-	bool			invoke_scheduler;
-	bool			disabled;
+	int32			invoke_scheduler;
 
 	// CPU topology information
 	int				topology_id[CPU_TOPOLOGY_LEVELS];