The 'sled_cpu:cpu_nsec' metric only reports values for the last 100 cpu_ids of the host

The `sled_cpu:cpu_nsec` metric only reports values for the last 100 cpu\_ids, e.g., on a gimlet cpus 28 -> 127, on a cosmo 156 -> 255 ``` # gimlet root@oxz_switch1:~# omdb oxql -e 'get sled_cpu:cpu_nsec | filter timestamp > @now() - 2h && sled_serial == "BRM42220077"' 2>/dev/null | grep cpu_id | awk '{print $2}' | sort -nu 28 29 30 ... 125 126 127 ``` ``` # cosmo support@oxz_switch0:~$ omdb oxql -e 'get sled_cpu:cpu_nsec | filter timestamp > @now() - 2h && sled_serial == "2PNX0JCM"' 2>/dev/null | grep cpu_id | awk '{print $2}' | sort -nu 156 157 158 ... 253 254 255 ``` This has been observed on dogfood, colo, and customer racks.