Move instance health check timeseries to TOML

[bnaecker]

No description provided.

[hawkw]

This looks correct to me!

Now that I'm actually looking at this, I kind of wonder if there's a way to combine this schema with the virtual_machine schema for PVPANIC etc (#6035). Would it make sense to have one virtual_machine schema that can query both the Nexus-emitted health check metrics and the Propolis-emitted PVPANIC metrics, and move stuff like Nexus IDs/addresses onto the check metrics themselves?

Naturally, I think it would make more sense to do that in a follow-up change, rather than here, as I think making the move to a TOML schema should probably be a commit that doesn't introduce any change in behavior, but I figured I'd mention it.

[bnaecker]

Thanks @hawkw! I do agree with you on both accounts -- it'd be nice to have one virtual_machine target which includes both the PVPANIC / vCPU stats and health checks together, and I would like to do that in a later commit. Doing it now would actually run afoul of the schema-consistency checks in oximeter as it collects all the data, since we're moving the source of those fields from a target to a metric!