Remove "obvious" deficiency in SSM scheduling when there are more than 10K open session #

When the Open Session number exceeds IBM's baseline of 10K, we can see Symphony SSM scheduling has an obvious "deficiency": Running task # starts to become much lower than the allocated slots to SSM bouncing between even 0 and very low; SSM busy on 1 thread (100%, while other threads pretty much idle). It is as if SSM were totally on haul. Note the total slots SSM can use is normally smaller the Open Session # often much smaller. The scheduling policy here is Proportional, although Priority policy also has this but likely less severe.

While we understand the 10K baseline, this deficiency is too obvious: Our sessions tend to be all at the highest possible say 10000. For a given # of allocated slots, SSM can only schedule certain # of highest priority sessions and any new or existing sessions do not need to be examined or re-examined for SSM dispatch scheduling E.g., if there's only 1 slot and it's used by a highest priority session, SSM doesn't need to do dispatch scheduling unil that 1 slot is freed. Sure this is not to say no scheduling related work such as preparation (like some sorting) needs to be done in the meantime but SSM doesn't need to be crazily busy as we observe about the scheduling thread

This looks like a relatively simple bottleneck to resolve instead of deep or fundamental change to SSM.

We can focus on proportional scheduling but good to see if similar bottleneck in priority scheduling.

Note it is very difficult to control the total open session # for our users/clients.