Tracking Issue for Managing CPU & RAM Capacity

[smklein]

See also: RFD 380

• Track provisioned CPU and RAM to services and instances on a per-sled basis in CRDB #2804
• Apply constraints to limit usage on a per-zone basis
• Use this information as a factor in sled selection for services and instances
• Add metrics to periodically monitor true CPU and RAM utilization by services within the sled

#1700 is a similar issue, but for disk space