Add sequencer timeout ereports

[mkeeter]

This PR adds two ereport flavors for Cosmo sequencer issues:

• hw.seq.regs is a raw dump of the registers, along with a reason why it was produced (MAPO, timeout, or "suddenly we're not in A0")
• hw.seq.timeout.* is a family of ereports representing timeouts at each stage of sequencing, along with a diagnosis. It includes an Option<u64> that refers to a hw.seq.regs ENA, so that we can also look at raw register values (if the diagnosis is not helpful).

For failure modes other than timeout, this PR doesn't send any additional ereports (just the hw.seq.regs value).

Consider this a starting point for brainstorming how we want to organize this data (e.g. one alternative is sending the high-level ereport first, then have the hw.seq.regs refer to it).

[hawkw]

take it or leave it: i think we could scrape off a couple bytes of ereport if this was:

seq_status: SeqStatusRegs

// ...

struct SeqStatusRegs {
   api: u32,
   raw: u32,
}

since it trades a couple bytes of nested K/V pairs for 11 bytes of repeating seq_ and _status for both fields?

[hawkw]

nit, take it or leave it: could save 2 bytes if this was

Suggested change

	debug_enables: u32,
	dbg_enables: u32,

[hawkw]

could save a few bytes if this was like

    rails: RailRegs,
    // ...
}
    
// ...

struct RailRegs {
    enables: u32,
    pgs: u32,
    pgs_max_hold: u32,
}

[hawkw]

part of me wants to say that these should be part of the ereport's class, but...i get why they're not. hm. 🤔