Cosmo sequencer PMBus monitoring ought to include more rails

[hawkw]

The other day we saw a Cosmo fail to sequence due to a timeout while waiting for the Group A rails. Looking at this sled's ringbuf we did not see POWER_GOOD on any of V1P5_RTC, V3P3_SP5_A1, or V1P8_SP5_A1 rails. The V3p3_SP5_A1 and V1P8_SP5_A1 rails are regulated by U116, which is an ISL68224. This is a PMBus part, and its PMBus alert pin is routed to the FPGA on PWR_CONT3_TO_FPGA1_ALERT_L, which can generate a corresponding interrupt to the SP.

Hubris could be enabling the PWR_CONT3_TO_FPGA_ALERT_L interrupt in the sequencer FPGA's Interrupt Enable Register (IER), and handle PMBus alerts from U116 similarly to how we handle alerts from the RAA229620A regulators for the VDDCR_CPU0_A0 and VDDCR_CPU1_A0 rails. It would be useful to be able to capture PMBus fault information from the ISL68224 as well.

Additionally, the current code for monitoring the V_core VRMs (the RAA229620As) presently only considers PMBus page 0 on those parts, which are the VDDCR_CPU0_A0 and VDDCR_CPU1_A0 rails. We probably ought to also be looking at page 1 on these VRMs as well, which corresponds to VDDCR_SOC_A0 and VDDIO_SP5_A0. After #2390 merges, we will clear faults on both pages when the PMBus alert is asserted, but we only look at the status registers and record potential faults from the page 0 rails.