Cosmo might need a little thermal loop tuning

When firing up a CPU synthetic load on a Group G (500W) processor on a Cosmo Rev B (with the good heatsink), the temperature on the processor overshoots a decent bit and then settles back down so we might need to be more aggressive in how we ramp the fans or what temperature we let the CPU idle at.

data from dashboard:

cosmo_b_500W_6.4T_drives_first_all_loadgen_then_cpu_loadgen_only.csv

As the filename implies, the first spike in CPU temperatures is from running

svcadm enable *loadgen*

The second spike is from running

svcadm enable cpu-loadgen

This is on the Cosmo Rev B behind niles which has a K.2 and an SP SWD. The SWD is setup with an alias of cosmo-b and you can login through the K.2 using ssh root@fe80::eaea:6aff:fe09:cafb%axf9 The only reason it has a K.2 is because it doesn't seem to want to advertise its host interface through the T6 currently

Start with the attached CSV and the dashboard data, then reproduce the two scenarios using svcadm enable *loadgen and svcadm enable cpu-loadgen on the Cosmo Rev B hardware. Locate the thermal-loop control entry point; the issue does not name a source file or test. Done means the reported CPU temperature overshoot and settling behavior have been addressed and verified against a comparable trace.